Content uploaded by Philip Wai Yan Chiu
Author content
All content in this area was uploaded by Philip Wai Yan Chiu on Mar 10, 2014
Content may be subject to copyright.
nature publishing group ORIGINAL CONTRIBUTIONS
STOMACH
84
The American Journal of GASTROENTEROLOGY VOLUME 105 | JANUARY 2010 www.amjgastro.com
INTRODUCTION
Although there is a general declining trend in the incidence
of peptic ulcer disease, peptic ulcer bleeding (PUB) remains
a major prevalent and clinically signi cant condition world-
wide. Recent studies suggest that the incidence of PUB ranges
from 20 to 60 per 100,000 population (1 – 3) , with an increasing
proportion of bleeding episodes related to the use of aspirin
and nonsteroidal anti-in ammatory drugs (NSAIDs). Despite
advances in endoscopic and pharmacological treatment in
PUB, the mortality associated with this condition stays at
around 5 – 10 % (4 – 10) . Although endoscopic therapy and adju-
vant use of proton pump inhibitors (PPIs) have been shown
to prevent recurrent bleeding, reduce the need for repeated
endoscopic therapy and surgery, and improve mortality, there
are still a substantial proportion of patients succumbing to the
bleeding episodes. One of the possible explanations is that PUB
is now predominantly a disease of the elderly, with over 60 %
of patients above the age of 60 years, and around 20 % over the
age of 80 years (11,12) . As elderly patients have more comorbid
illness, are more likely users of aspirin and NSAIDs, and are
less tolerant of hemodynamic insult, the management of this
high-risk population of PUB patient is a major challenge.
Causes of Mortality in Patients With Peptic Ulcer
Bleeding: A Prospective Cohort Study of 10,428 Cases
Joseph J.Y. Sung , MD, PhD, FRCP 1 , K e l v i n K . F . T s o i , P h D 1 , T e r r y K . W . M a , M B C h B 1 , M a n - Y e e Yu n g , B N 1 , J a m e s Y . W. L a u , M D , F R C S 1 a n d
P h i l i p W . Y . C h i u , M D , F R C S 1
OBJECTIVES: Despite advances in endoscopic and pharmacological treatment for peptic ulcer bleeding (PUB),
mortality remains at 5 – 10 % worldwide. Our aim was to investigate the causes of death in a
prospective cohort of PUB in a tertiary referral center.
METHODS: Between 1993 and 2005, all patients with upper gastrointestinal bleeding (UGIB) admitted to
the Prince of Wales Hospital were prospectively registered. Demographic data, characteristics of
ulcer, and pharmacological, endoscopic, and surgical therapy, were documented. Mortality cases
were classifi ed as (A) bleeding-related death (A1: uncontrolled bleeding, A2: within 48 h after
endoscopy, A3: during surgery for uncontrolled bleeding, A4: surgical complications or within
1 month after surgery, and A5: endoscopic related mortality) or (B) non-bleeding-related death
(B1: cardiac causes, B2: pulmonary causes, B3: cerebrovascular disease, B4: multiorgan failure,
and B5: terminal malignancy).
RESULTS: In all, 18,508 cases of UGIB were enrolled; among them, 10,428 cases from 9,375 patients
were confi rmed to have PUB, and 577 (6.2 % ) patients died. There were signifi cantly more
patients who died of non-ulcer bleeding causes (79.7 % ) than bleeding causes (18.4 % ). The
mean (s.d.) age of those who died of bleeding-related causes was higher (75.4 (12.6) years)
than that of those who died of non-bleeding causes (71.7 (13.1) years) ( P = 0.010). Most
bleeding-related deaths occurred when immediate control of bleeding failed (29.2 % ) or when
patients died within 48 h after endoscopic therapy (25.5 % ). Among those who died of
non-bleeding-related causes, multiorgan failure (23.9 % ), pulmonary conditions (23.5 % ), and
terminal malignancy (33.7 % ) were most common.
CONCLUSIONS: The majority of PUB patients died of non-bleeding-related causes. Optimization of management
should aim at reducing the risk of multiorgan failure and cardiopulmonary death instead of
focusing merely on successful hemostasis.
Am J Gastroenterol 2010; 105:84–89; doi:10.1038/ajg.2009.507; published online 15 September 2009
1 Institute of Digestive Disease, The Chinese University of Hong Kong , Shatin , NT , Hong Kong . Correspondence: Joseph J.Y. Sung, MD, PhD, FRCP , Department of
Medicine and Therapeutics, Chinese University of Hong Kong, Prince of Wales Hospital , Shatin , NT , Hong Kong . E-mail: joesung@cuhk.edu.hk
Received 12 January 2009; accepted 23 April 2009
see related editorial on page 90
© 2010 by the American College of Gastroenterology The American Journal of GASTROENTEROLOGY
85
STOMACH
Causes of Mortality in Patients With PUB
Most studies in PUB intervention use recurrent bleeding,
repeated endoscopic therapy, surgery, and blood transfusion
as the outcome parameter. Yet very few have shown improve-
ment in the survival of patients despite a reduction in the recur-
rent bleeding rate. ere seems to be a discrepant outcome in
endoscopic end points vs. clinical end points. As the ultimate
aim of treatment of gastrointestinal bleeding is to save lives, it is
important to elucidate the causes of death in these patients. We
conducted this study to investigate the cause of death in PUB
patients in a large cohort from a tertiary referral center.
METHODS
e Prince of Wales Hospital, a teaching hospital for the Chinese
University of Hong Kong, serves a population of 1.2 million
people in the northeastern territory of the Hong Kong SAR.
An upper gastrointestinal bleeding (UGIB) registry was estab-
lished to collect demographic information, characteristics of
bleeding pathologies, and pharmacological, endoscopic, or
surgical treatments for peptic ulcer patients with hematem-
esis or melena since 1993. Findings at index endoscopy on
admission and at the endoscopic therapy of each patient were
recorded by endoscopists. Follow-up hemostatic procedures
were documented if patients required second endoscopy for
rebleeding ulcer within the same admission. Patients admit-
ted for PUB or those who developed bleeding a er admis-
sion for other medical conditions were prospectively collected
between 1993 and 2005. In this study, we focus on PUB. Data
extracted from the UGIB registry included ulcer location, stig-
mata of recent hemorrhage (according to the Forrest classi -
cation), nature and number of comorbid illness, Helicobacter
pylori infection, as well as the use of H
2 receptor antagonists or
PPIs, NSAIDs (including conventional NSAID and cyclooxy-
genase 2-speci c inhibitors), or antiplatelet agents (includ-
ing aspirin and clopidogrel). Presentation of hemodynamic
status, requirement of endoscopic treatment, and occurrence
of rebleeding was recorded. All clinical details were based on
the initial records of admission.
Mortality within 30 days (30-day mortality) a er index bleed-
ing is the primary outcome of this study. ese are counted as
any causes of death within 30 days a er index bleeding, includ-
ing patients who died while still in hospital or those who had
been discharged from hospital. A er discharge from hospital,
patients with PUB were scheduled for follow-up in the ulcer
clinic at 4 – 6 weeks. Mortality cases within 30 days a er dis-
charge were identi ed by (i) those who were subsequently read-
mitted to hospital and died or (ii) by calling patients at home
for failing to turn up at follow-up visits. In-hospital mortality
is de ned as all cases of death during the period of hospital
stay irrespective of the duration of hospitalization. erefore,
this included those patients who died 30 days a er admission.
Although the majority of cases were admitted for symptoms
of UGIB, there were also cases that were admitted for other
medical conditions but developed UGIB during the course of
hospital stay.
Comorbidity data were based on (i) detailed clinical his-
tory on admission, (ii) electronic patient record in the hospital
database, and (iii) structured questionnaire a er endoscopy. It
was presented as a simple dichotomous question by dedicated
research sta of the Endoscopy Center. All data were entered
prospectively. e de nitions of serious comorbid illness
included decompensated cardiac, pulmonary, hepatic, neuro-
logical and renal disease, and advanced malignancy.
Causes of mortality were classi ed into (A) bleeding-related
death and (B) non-bleeding-related death. Under the category
of bleeding-related death, cases were further subcategorized
into (A1) uncontrolled bleeding, (A2) within 48 h a er endo-
scopy, (A3) during surgery for uncontrolled bleeding, (A4) sur-
gical complications or within 1 month a er surgery, and (A5)
endoscopic-related mortality. Under non-bleeding-related
death, cases were subcategorized into (B1) cardiac causes, (B2)
pulmonary causes, (B3) cerebrovascular disease, (B4) multi-
organ failure, and (B5) terminal malignancy. Mortality was
registered in the system in Hospital Authority. Causes of death
were independently determined by two clinicians (T.K.W.M.
and P.W.Y.C.) on the basis of clinical records in UGIB registry
and the death certi cates. When discrepancies were found, a
consensus agreement would be made with the third investigator
(J.J.Y.S.) by retrieving the complete hospital record for review.
Univariate analyses were carried out using SPSS (version
13.0; SPSS, Chicago, IL). Di erences in categorical variables
were evaluated using Pearson ’ s 2 test or Fisher ’ s exact test when
appropriate. Di erences in continuous variables were evaluated
using Student ’ s t -test for independent samples, a er verifying
homogeneity of variance with Levene ’ s test. e bleeding-related
mortalities each year were plotted against time and 2 test for
trend was carried out to detect if there is a consistent inclin-
ing or declining trend. Any missing record was excluded dur-
ing the statistical calculation. Subgroup analyses were carried
out for the patients admitted for gastrointestinal bleeding. All
tests were performed two-sided and considered to be statisti-
cally signi cant for P value < 0.05. Logistic regression was used
to estimate the odds ratio (OR) and the 95 % con dence inter-
val (CI) of various possible risk factors for the bleeding-related
mortality. Risk factors with a P value of < 0.10 in univariate
analyses were included in a multiple logistic regression model
and analyzed using the backward approach. e 95 % CI of
the OR was used for assessing the statistical signi cance at the
conventional level of 0.05.
RESULTS
Between 1993 and 2005, a total of 18,508 UGIB episodes were
entered into the GIB registry. Among them, a total of 10,428
cases from 9,375 patients were con rmed to have PUB. is
included 612 (6.5 % ) patients who died during hospital stay or
within 30 days a er index endoscopy, 8,667 (92.5 % ) patients
who survived the bleeding episode(s), and 96 (1 % ) dis-
charged patients who died a er 30 days of index endoscopy.
A total of 577 (6.2 % ) patients died within 30 days a er index
The American Journal of GASTROENTEROLOGY VOLUME 105 | JANUARY 2010 www.amjgastro.com
86
STOMACH
Sung et al.
endoscopy. e characteristics of patients were compared
between those who survived and those who died within 30
days a er endoscopy ( Table 1 ). e surviving patients were
younger (61.0 vs. 72.5 years), with less severe comorbid illness.
ey were more likely to have a previous history of ulcer and
H. pylori infection, and less likely to be users of H
2 blocker / PPI
and NSAID / aspirin. ey were more likely to be admitted just
for UGIB and were found to have clean-based ulcer. Patients
with hemodynamic shock and clinical rebleeding were more
likely to die within 40 days a er index bleeding.
Out of 577 mortality cases within 30 days, 106 (18.4 % )
patients were classi ed as bleeding-related mortality and 460
(79.7 % ) as non-bleeding-related mortality ( Tabl e 2 ). Causes
of 30-day mortality in 11 (1.9 % ) patients were unclassi-
ed because of insu cient clinical information availability
( Tabl e 2 ). A unanimous decision on the cause of death was
reached in 551 (95.5 % ) cases in the rst round. Consen-
sus diagnoses reached a er clinical notes review and a third
clinician ’ s input were given in 4.5 % .
e mean (s.d.) age of 106 patients who died of bleeding-
related mortality within 30 days was 75.4 (12.6) years, of which
60.4 % were males. Among these patients, uncontrolled or
recurrent bleeding from peptic ulcer was reported in 29.2 % ,
and death from circulatory failure within 48 h a er endoscopy
was reported in 25.5 % . Surgical or endoscopic complications
accounted for 29.2 and 13.2 % , respectively, of cases of 30-day
mortality. Only 2.8 % of subjects died as a result of uncontrolled
bleeding during surgery ( Ta ble 2 ).
e mean (s.d.) age of 460 patients who died of non-bleeding-
related causes was 71.7 (13.1) years, of which 65.9 % were males.
Among these patients, the most common causes of death were
terminal malignancy (33.7 % ), multiorgan failure (23.9 % ), and
pulmonary disease (23.5 % ). Cardiac diseases (including acute
coronary syndrome and heart failure) and cerebrovascular
diseases constituted 13.5 and 5.4 % , respectively. As a whole
group, 435 of 577 (75.4 % ) of 30-day mortality cases were due to
terminal malignancy, multiorgan failure, and cardiopulmonary
decompensation.
A total of 469 (4.5 % ) patients died during the period of
hospital stay a er the index bleeding. Among these patients,
104 (22 % ) died of bleeding-related causes, 358 (77 % ) died of
non-bleeding-related causes, and 7 (1 % ) were unclassi ed.
Overall, 50.0 % in the bleeding-related group were attributed
to uncontrolled bleeding (24.0 % ) or circulatory failure with
48 h a er endoscopy (26.0 % ), and 80.7 % in the non-bleeding-
related group were attributed to terminal malignancy (23.7 % ),
multiorgan failure (29.6 % ), and pulmonary diseases (27.4 % ).
Patients who had PUB before admission and those who
developed PUB a er admission are listed separately in Tabl e 3 .
Patients who had PUB before admission were older and were
more likely to have a history of peptic ulcers and a di erent
distribution of peptic ulcers.
Univariate analysis showed that patients who died of bleed-
ing-related causes were older than those who died of non-
bleeding-related causes ( P = 0.010) ( Tabl e 4 ). ose who had
a history of peptic ulcer disease were also more likely to die
from bleeding than those who had no ulcer history (31.1 vs.
18.5 % , P = 0.005). e use of NSAID or aspirin is signi cantly
more common in the bleeding-related death group than in
the non-bleeding death group (86.4 vs. 63.2 % , P = 0.003). e
anatomical locations of peptic ulcers (i.e., gastric, duodenal,
or anastomotic ulcer) had no bearing on the cause of death.
On the other hand, patients admitted with hemodynamic
shock (39.4 vs. 22.4 % , P < 0.001) and those with actively bleed-
ing ulcers on index endoscopy (42.5 vs. 22.4 % , P < 0.001)
were associated with a higher risk of bleeding-related death. e
Table 1 . Characteristics of patients who survived or died from
peptic ulcer bleeding
Surviving
patients
( n =8,667)
Patients died
within 30 days
after endo scopy
( n =577) P value
Mean age, years (s.d.) 61.0 (18.4) 72.5 (13.1) < 0.001
Male ( % ) 5,715 (65.9) 372 (64.5) 0.471
Patients with ulcer
history ( % )
3,053 (35.3) 118 (20.6) < 0.001
Ulcer location ( % )
Gastric ulcer 3,204 (37.0) 245 (42.5)
Duodenal ulcer 4,579 (52.8) 242 (41.9)
Both 655 (7.6) 63 (10.9)
Anastomotic ulcer 227 (2.6) 27 (4.7) < 0.001
Severe comorbid illness ( % )
0 5,861 (67.6) 123 (21.3)
1 2,106 (24.3) 287 (49.7)
2 624 (7.2)
139 (24.1)
≥ 3 76 (0.9) 28 (4.9) < 0.001
Use of H
2 blocker
or PPI ( % )
654 (7.9) 66 (12.2) < 0.001
Use of NSAID or
aspirin ( % )
1385 (48.7) 184 (67.4) < 0.001
Helicobacter pylori
infection ( % )
3,521 (61.0) 65 (23.0) < 0.001
Type of ulcer ( % )
Active bleeding ulcer 1,362 (17.1) 149 (25.8)
Clot or vessel at base
of ulcer
2,323 (29.1) 211 (36.6)
Clean-based ulcer 4,301 (53.9) 217 (37.6) < 0.001
Hemodynamic shock
( % )
718 (8.4) 144 (25.2) < 0.001
Admitted for UGIB ( % ) 7,888 (91.0) 334 (58.0) < 0.001
Clinical rebleeding ( % ) 233 (8.8) 73 (27.8) < 0.001
NSAID, nonsteroidal anti-infl ammatory drug; PPI, proton pump inhibitor; UGIB,
upper gastrointestinal bleeding.
© 2010 by the American College of Gastroenterology The American Journal of GASTROENTEROLOGY
87
STOMACH
Causes of Mortality in Patients With PUB
percentage of patients admitted for UGIB, patients with clinical
evidence of recurrent bleeding, and patients requiring repeated
endoscopic therapy were not signi cantly di erent between
the two groups of 30-day mortality cases ( Tab le 4 ). In non-
bleeding-related mortality, a signi cantly higher proportion
of patients with severe comorbid illnesses have been reported
(63.2 vs. 82.6 % , P < 0.001). H. pylori infection of the two groups
showed no signi cant di erence ( Tab le 4 ). Using a multivariate
logistic regression model, use of NSAID or aspirin, as well as
the endoscopic appearance of ulcer and hemodynamic shock,
was found to be an independent variable predicting bleeding-
related death ( Tabl e 5 ).
Overall, 50.0 % of patients in the bleeding-related mortality
group and 28.9 % patients in the non-bleeding-related mor-
tality group died within 3 days of index bleeding ( P < 0.001).
Morta lity related to bleeding over the period 1993 – 2005 ranges
from 13.3 to 39.4 % ( Figure 1 ). ere is no signi cant trend
shown on the bleeding-related mortality in the study period
( P = 0.087).
DISCUSSION
Mortality remains high in patients presenting with PUB and
this study shows that around 80 % of them died of non-bleed-
ing-related causes. In this prospective cohort study enrolling
over 10,000 cases of PUB, terminal malignancy, multiorgan
failure, and cardiopulmonary decompensation were the most
important causes of death, accounting for 75 % of cases.
e ndings from this study are strengthened by the fact that
in this single referral center a dedicated team of medical and
surgical experts provided service to PUB patients. e protocol
of management has not been signi cantly changed in the last 15
Table 2 . Causes of death of 577 patients who died within
30 days after index bleeding
Mortality cases
n =577 Subcategories n Percentage
Bleeding related
n =106 (18.4 % )
Uncontrolled bleeding /
rebleeding
31 29.2
Within 48 h after endo-
scopy without other cause
27 25.5
During surgery for
uncontrolled bleeding
3 2.8
Surgical complications or
within 1 month after sur-
gery without other cause
31 29.2
Endoscopy-related
complication
14 13.2
Non – bleeding
related n =460
(79.7 % )
Cardiac diseases
(e.g., ACS, heart failure)
62 13.5
Pulmonary diseases
(e.g., COPD, pneumonia)
108 23.5
Multiorgan failure 110 23.9
Neurological diseases
(e.g., stroke)
25 5.4
Terminal malignancy 155 33.7
Unclassifi ed
n
=11 (1.9 % )
ACS, acute coronary syndrome; COPD, chronic obstructive pulmonary disease.
Table 3 . Comparison of patients with PUB as a presenting
symptom before admission or after patients were admitted for
other reasons (excluding 11 cases with unclassifi ed cause of
death)
PUB before
admission
( n =329)
PUB after
admission
( n =237) P value
Mean age, years (s.d.) 73.5 (12.7) 70.9 (13.6) 0.021
Male ( % ) 208 (63.2) 159 (67.1) 0.342
Patients with ulcer
history ( % )
80 (24.5) 37 (15.7) 0.011
Ulcer location ( % )
Gastric ulcer 148 (45.0) 92 (39.0)
Duodenal ulcer 122 (37.1) 115 (48.3)
Both 43 (13.1) 20 (8.5)
Anastomotic ulcer 16 (4.9) 10 (4.2) 0.040
Severe comorbid illness ( % )
0 80 (24.3) 39 (16.5)
1 161 (48.9) 120 (50.8)
2 73 (22.2) 65 (27.5)
≥ 3 15 (4.6) 13 (5.5) 0.116
Use of H
2 blocker or
PPI ( % )
35 (11.5) 31 (13.8) 0.436
Use of NSAID or aspirin
( % )
106 (68.4) 73 (65.2) 0.582
Helicobacter pylori
infection ( % )
38 (23.8) 26 (22.6) 0.825
Type of ulcer ( % )
Active bleeding ulcer 90 (27.4) 58 (24.6)
Clot or vessel at base
of ulcer
125 (38.0) 83 (35.2)
Clean-based ulcer 114 (34.7) 96 (40.5) 0.360
Hemodynamic shock ( % ) 80 (24.5) 63 (27.0) 0.491
Clinical rebleeding ( % ) 31 (21.5) 27 (27.8) 0.261
Required endoscopic
re-treatment ( % )
23 (16.0) 22 (22.7) 0.190
Bleeding-related
mortality ( % )
65 (19.8) 41 (17.3) 0.460
NSAID, nonsteroidal anti-infl ammatory drug; PPI, proton pump inhibitor; PUB,
peptic ulcer bleeding.
The American Journal of GASTROENTEROLOGY VOLUME 105 | JANUARY 2010 www.amjgastro.com
88
STOMACH
Sung et al.
using a heater probe. Hemoclips were used but only in < 5 %
of the cases. In the last 8 years, intravenous high-dose PPIs
(including omeprazole, pantoprazole, and esomeprazole) had
been added as an adjuvant to endoscopic therapy. A er endo-
scopic therapy, patients were nursed in a closely monitored
gastrointestinal bed in the combined unit. Oral feeding was
resumed on the next day or as soon as the patients ’ condition
became stable. Intravenous PPIs were maintained for 3 days
and replaced by oral medications including anti-helicobacter
therapy, if deemed appropriate. Indications for surgery had
remained unchanged, i.e., for uncontrolled bleeding in index
endoscopy or recurrent bleeding not responding to endoscopic
therapy. Patients would be nursed in the intensive care unit if
they required, or if they were likely to require ventilatory sup-
port or cardiac resuscitation. Variations in management proto-
col, suboptimal endoscopic or medical therapy, and decisions
on surgery or intensive care unit would not confound the nd-
ings in this study.
e implications of this nding are self-evident. If morta lity
among PUB cases is not necessarily related to uncontrolled
bleeding, it is very hard to improve on it through advance-
ment in treatment targeting on bleeding control alone. e
rebleeding rates from major studies in PUB, using a combina-
tion of endoscopic hemostasis and adjuvant acid-suppressing
therapy, have been shown to be reduced to 5 – 8 % (13,14) and it
years. Endoscopy has always been the rst line of investigation
and treatment of PUB and has been o ered within 24 h a er
admission. Endoscopic hemostasis has always been a combina-
tion of epinephrine injection followed by thermal coagulation
Table 4 . Univariate analysis between bleeding and non-
bleeding-related mortality (excluding 11 cases with
unclassifi ed cause of death)
Bleeding
related mortality
( n =106)
Non-bleeding-
related mortality
( n =460) P value
Mean age (s.d.) 75.4 (12.6) 71.7 (13.1) 0.010
Male ( % ) 64 (60.4) 303 (65.9) 0.286
Patients with ulcer
history ( % )
32 (31.1) 85 (18.5) 0.005
Ulcer location ( % )
Gastric ulcer 35 (33.0) 205 (44.6)
Duodenal ulcer 54 (50.9) 183 (39.8)
Both 11 (10.4) 52 (11.3)
Anastomotic
ulcer
6 (5.7) 20 (4.3) 0.129
Severe comorbid illness ( % )
0 39 (36.8) 80 (17.4)
1 40 (37.7) 241 (52.4)
2 21 (19.8) 117 (25.4)
≥ 3 6 (5.7) 22 (4.8)
< 0.001
Use of acid sup-
pressing agents ( % )
11 (11.3) 55 (12.7) 0.713
Use of NSAID or
aspirin ( % )
38 (86.4) 141 (63.2) 0.003
Helicobacter pylori
infection ( % )
15 (31.9) 49 (21.5) 0.124
Endoscopic appearance of ulcer ( % )
Active bleeding
ulcer
45 (42.5) 103 (22.4)
Clot or vessel at
base of ulcer
46 (43.4) 162 (35.2)
Clean-based
ulcer
15 (14.2) 195 (42.4) < 0.001
Hemodynamic
shock ( % )
41 (39.4) 102 (22.4) < 0.001
Admitted for UGIB
( % )
65 (61.3) 264 (57.5) 0.474
Clinical rebleeding
( % )
16 (27.6) 42 (23.0) 0.472
Required endo-
scopic re-treatment
( % )
14 (24.1) 31 (16.9)
0.220
NSAID, nonsteroidal anti-infl ammatory drug; UGIB, upper gastrointestinal
bleeding.
Table 5 . Multivariate analysis between bleeding and
non-bleeding-related mortality (excluding 11 cases with
unclassifi ed cause of death)
OR 95 % CI
Use of NSAID or aspirin 3.70 1.43 – 9.60
Endoscopic appearance of ulcer
Active bleeding ulcer 12.96 2.82 – 59.55
Clot or vessel at base of ulcer 12.29 2.75 – 54.95
Hemodynamic shock 3.75 1.77 – 7.96
CI, confi dence interval; NSAID, nonsteroidal anti-infl ammatory drug; OR, odds ratio.
Bleeding mortality
Non-bleeding
mortality
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005
Year
Deaths (%)
Figure 1 . Incidences of bleeding-related and non-bleeding-related
mortality between the years 1993 – 2005.
© 2010 by the American College of Gastroenterology The American Journal of GASTROENTEROLOGY
89
STOMACH
Causes of Mortality in Patients With PUB
would be very di cult, if not impossible, to reduce this further
to approach zero. As the majority of patients with PUB these
days are of advanced age, users of NSAID and aspirin, impli-
cated with multiple comorbid illnesses, the endeavor to further
reduce this rebleeding rate is an uphill battle. Energy should be
directed to improve on supportive care or prevent complica-
tions of the other organ systems instead of focusing merely on
the gastrointestinal tract.
It is common practice to withhold aspirin or clopidogrel for a
period of time a er PUB. Discontinuation of antiplatelet agents
may have increased the risk of non-bleeding-related deaths as
patients with cardiovascular and cerebrovascular diseases were
exposed to unprotected vascular complications. In our earlier
study comparing the early resumption of aspirin vs. withholding
for 8 weeks a er PUB, there was a signi cantly increased mortality
in the latter group (15) . e risk and bene t of early resumption
of antiplatelet agents deserve further evaluation. Future studies
should look at the management of PUB patient as a whole instead
of focusing only on gastrointestinal bleeding. On the other hand,
gastrointestinal bleeding can be seen as a terminal event in some
patients who are dying from other conditions such as metastatic
cancers and multiorgan failure. It would be very hard to make any
substantial improvement in their clinical outcome.
ere are several limitations in this study. First, as a single
center, the generalizability of this nding is a concern. However,
the prospective collection of these data through a well-organ-
ized GIB registry guarantees good quality of information for
analysis. e high volume of PUB cases seen by this center also
strengthens the ndings in this study. In multicenter studies
or studies using retrospective data, the variable quality of data
may impose di culties in interpretation. Second, the deci-
sion of cause of death may not be strictly forward. ere were
cases with evidence on the cause of death pointing in di erent
directions and, indeed, in some cases, more than one possible
cause of death was possible. To minimize bias, two experienced
clinicians made their diagnosis independently and a third clini-
cian ’ s opinion was added in an attempt to reach a consensus. e
unanimous diagnosis reached by 95 % in the rst round suggests
that a majority of the cases were undisputed.
In conclusion, a majority of patients with PUB in hospital died
of non-bleeding-related causes. is nding calls for a stronger
support of other major organ systems in these patients. Future
studies should be directed to explore ways to improve the over-
all outcome of the patients.
CONFLICT OF INTEREST
Guarantor of the article: Joseph J.Y. Sung, MD, PhD, FRCP.
Speci c author contributions: Study design and writing of
the paper: Joseph J.Y. Sung; result analysis: Kelvin K.F. Tsoi;
data collection: Terry K.W. Ma and Man-Yee Yung; and data
collection and writing of the paper: James Y.W. Lau and
Philips W.Y. Chiu.
Financial support: None.
Potential competing interests: Joseph J.Y. Sung is on the
advisory board of AstraZeneca and is a speaker for Astra-
Zeneca and Nycomed. James Y.W. Lau is on the advisory
boards of and is a speaker for AstraZeneca and Nycomed.
Study Highlights
WHAT IS CURRENT KNOWLEDGE
3 Peptic ulcer bleeding has a mortality rate of 5 – 10 % .
3 Acid-suppressing agents, endoscopic hemostasis, and
surgery are effective in bleeding control.
3 There is little improvement in survival rate despite these
advances in treatment.
WHAT IS NEW HERE
3 Overall, 80 % of mortality is not related to bleeding.
3 Multiorgan failure, cardiopulmonary conditions, and
terminal malignancy constitute the most common
causes in these patients.
3 Over 50 % of bleeding-related mortality occurred within
3 days after onset of bleeding.
REFERENCES
1 . L a s s e n A , H a l l a s J , S c h a alitzky de Muckadell OB . Complicated and
uncomplicated peptic ulcers in a Danish county 1993 – 2002: a population-
based cohort study . Am J Gastroenterol 2006 ; 101 : 945 – 53 .
2 . L e w i s J D , B i l k e r W B , B r e n s i n g e r C et al. Hospitalization and mortality rates
from peptic ulcer disease and GI bleeding in the 1990s: relationship to sales
of nonsteroidal anti-in ammatory drugs and acid suppression medications .
Am J Gastroenterol 2002 ; 97 : 2540 – 9 .
3 . R a m s o e k h D , v a n L e e r d a m M E , R a u w s E A et al. Outcome of peptic ulcer
bleeding, nonsteroidal anti-in ammatory drug use, and Helicobacter pylori
infection . Clin Gastroenterol Hepatol 2005 ; 3 : 859 – 64 .
4 . C o o p e r G S , Y u a n Z , R o s e n t h a l G E et al. Lack of gender and racial di er-
ences in surgery and mortality in hospitalized medicare bene ciaries with
bleeding peptic ulcer . J Gen Intern Med 1997 ; 12 : 485 – 90 .
5 . K u b b a A K , C h o u d a r i C , R a j g o p a l C et al. Reduced long-term survival fol-
lowing major peptic ulcer haemorrhage . Br J Surg 1997 ; 84 : 265 – 8 .
6 . H a s s e l g r e n G , C a r l s s o n J , L i n d T et al. Risk factors for rebleeding and fatal
outcome in elderly patients with acute peptic ulcer bleeding . Eur J Gastro-
enterol Hepatol 1998 ; 10 : 667 – 72 .
7 . Bini EJ , Cohen J . Endoscopic treatment compared with medical therapy
for the prevention of recurrent ulcer hemorrhage in patients with adherent
clots . Gastrointest Endosc 2003 ; 58 : 707 – 14 .
8 . C h i u P W Y , L a m C Y , L e e S W et al. E ect of scheduled second therapeutic
endoscopy on peptic ulcer rebleeding: a prospective randomized trial .
Gut 2003 ; 52 : 1403 – 7 .
9 . M o s e H , L a r s e n M , R i i s A et al. irty-day mortality a er peptic ulcer
bleeding in hospitalized patients receiving low-dose aspirin at time of
admission . Am J Geriatr Pharmacother 2006 ; 4 : 244 – 50 .
1 0 . o m s e n R W , R i i s A , C h r i s t e n s e n S et al. Diabetes and 30-day mortality
from peptic ulcer bleeding and perforation . Diabetes Care 2006 ; 29 : 805 – 10 .
11 . Rockall TA , Logan RF , Devlin HB et al. Incidence of and mortality from
acute upper gastrointestinal haemorrhage in the United Kingdom. Steering
Committee and members of the National Audit of Acute Upper Gastro-
intestinal Haemorrhage . BMJ 1995 ; 311 : 222 – 6 .
1 2 . O h m a n n C , I m h o f M , R u p p e r t C et al. Time-trends in the epidemiology of
peptic ulcer bleeding . Scand J Gastroenterol 2005 ; 40 : 914 – 20 .
1 3 . L a u J Y , S u n g J J , L e e K K et al. E ect of intravenous omeprazole on
recurrent bleeding a er endoscopic treatment of bleeding peptic ulcers .
N Engl J Med 2000 ; 343 : 310 – 6 .
1 4 . S u n g J J , T s o i K K , L a i L H et al. Endoscopic clipping vs. injection and
thermo-coagulation in the treatment of bleeding non-variceal upper
gastrointestinal bleeding: a meta-analysis . Gut 2007 ; 56 : 1364 – 72 .
1 5 . S u n g J J , L a u J Y , C h i n g J Y et al. Can aspirin be reintroduced with proton
pump inhibitor infusion a er endoscopic hemostasis? A double blinded
randomized controlled trial . Gastroenterology 2006 ; 130 (Suppl 2) : A44 .
