Annals of Internal Medicine
Continuation of Low-Dose Aspirin Therapy in Peptic Ulcer Bleeding
A Randomized Trial
Joseph J.Y. Sung, MD, PhD; James Y.W. Lau, MD; Jessica Y.L. Ching, MPH; Justin C.Y. Wu, MD; Yuk T. Lee, MD; Philip W.Y. Chiu, MD;
Vincent K.S. Leung, MD; Vincent W.S. Wong, MD; and Francis K.L. Chan, MD
Background: It is uncertain whether aspirin therapy should be
continued after endoscopic hemostatic therapy in patients who
develop peptic ulcer bleeding while receiving low-dose aspirin.
Objective: To test that continuing aspirin therapy with proton-
pump inhibitors after endoscopic control of ulcer bleeding was not
inferior to stopping aspirin therapy, in terms of recurrent ulcer
bleeding in adults with cardiovascular or cerebrovascular diseases.
Design: A parallel randomized, placebo-controlled noninferiority
trial, in which both patients and clinicians were blinded to treatment
assignment, was conducted from 2003 to 2006 by using computer-
generated numbers in concealed envelopes. (ClinicalTrials.gov reg-
istration number: NCT00153725)
Setting: A tertiary endoscopy center.
Patients: Low-dose aspirin recipients with peptic ulcer bleeding.
Intervention: 78 patients received aspirin, 80 mg/d, and 78 re-
ceived placebo for 8 weeks immediately after endoscopic therapy.
All patients received a 72-hour infusion of pantoprazole followed
by oral pantoprazole. All patients completed follow-up.
Measurements: The primary end point was recurrent ulcer bleed-
ing within 30 days confirmed by endoscopy. Secondary end points
were all-cause and specific-cause mortality in 8 weeks.
Results: 156 patients were included in an intention-to-treat anal-
ysis. Three patients withdrew from the trial before finishing follow-
A spirin has been increasingly used to treat cardiovas-
cular and cerebrovascular diseases, especially in the
aging population, yet it causes a 2- to 3-fold increase in
the risk for dose-related peptic ulcer bleeding (1–3).
Enteric-coated and buffered aspirin do not seem to be
safer than a similar dose of plain aspirin (4). A meta-
analysis that pooled data from 24 randomized studies
suggests that risk for aspirin-induced ulcer bleeding is
not reduced with long-term use of the drug (5). Toler-
ance to gastropathic effects of aspirin cannot be substan-
tiated. Aspirin-induced ulcers often exist without symp-
toms of dyspepsia (6). Elderly patients who test positive
for Helicobacter pylori infection have added risk for pep-
tic ulcer disease and complications (7). The risk for
upper gastrointestinal bleeding is similar for recipients
of aspirin and nonaspirin antiplatelet agents, such as
clopidogrel (8). Patients who receive a combination of
antiplatelet agents are at higher risk for gastrointestinal
bleeding (2, 9). Antisecretory therapies, including proton-
pump inhibitors and H2-receptor antagonists, confer
protection against upper gastrointestinal bleeding in an-
tiplatelet recipients; the reduction in risk is greater with
proton-pump inhibitors (10).
Article
up. Recurrent ulcer bleeding within 30 days was 10.3% in the
aspirin group and 5.4% in the placebo group (difference, 4.9
percentage points [95% CI, ⫺3.6 to 13.4 percentage points]).
Patients who received aspirin had lower all-cause mortality rates
than patients who received placebo (1.3% vs. 12.9%; difference,
11.6 percentage points [CI, 3.7 to 19.5 percentage points]). Pa-
tients in the aspirin group had lower mortality rates attributable to
cardiovascular, cerebrovascular, or gastrointestinal complications
than patients in the placebo group (1.3% vs. 10.3%; difference, 9
percentage points [CI, 1.7 to 16.3 percentage points]).
Limitations: The sample size is relatively small, and only low-dose
aspirin, 80 mg, was used. Two patients with recurrent bleeding in
the placebo group did not have further endoscopy.
Conclusion: Among low-dose aspirin recipients who had peptic
ulcer bleeding, continuous aspirin therapy may increase the risk for
recurrent bleeding but potentially reduces mortality rates. Larger
trials are needed to confirm these findings.
Primary Funding Source: Institute of Digestive Disease, Chinese
University of Hong Kong.
Ann Intern Med. 2010;152:1-9. www.annals.org
For author affiliations, see end of text.
This article was published at www.annals.org on 1 December 2009.
When patients present with peptic ulcer bleeding, the
usual protocol is to treat the active bleeding with an endo-
scopic device, offer antisecretory therapy, and discontinue
aspirin or other antiplatelet agents until the ulcer heals.
However, there is a risk for cardiovascular and cerebrovas-
cular events and death when antiplatelet agents are discon-
tinued. The balance of risk and benefit for prescribing anti-
platelet agents under these situations is an estimation of the
chance of developing recurrent upper gastrointestinal
bleeding against the chance of vascular thrombotic events
in the cardiac and neurologic systems (11). We previously
showed that intravenous infusion of proton-pump inhibi-
See also:
Print
Editors’ Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Editorial comment. . . . . . . . . . . . . . . . . . . . . . . . . . . 52
Summary for Patients. . . . . . . . . . . . . . . . . . . . . . . I-20
Web-Only
Conversion of graphics into slides
© 2010 American College of Physicians 1
Article Continuation of Low-Dose Aspirin in Peptic Ulcer Bleeding
Context
What happens if patients who take aspirin to prevent
cardiovascular disease continue to take it after an acute
gastrointestinal bleeding event?
Contribution
This trial included 156 adults with cardiovascular disease,
history of aspirin use, and acute peptic ulcer bleeding. Im-
mediately after successful endoscopic treatment, they were
randomly assigned to receive low-dose aspirin (80 mg/d)
or placebo for 8 weeks. All patients also received panto-
prazole. More aspirin recipients than placebo recipients
(10% vs. 5%) had recurrent ulcer bleeding within 30
days, although fewer aspirin recipients died (1% vs. 13%).
Caution
The study was small. Some deaths in the placebo group
were from causes not normally prevented by aspirin.
—The Editors
tors reduced the incidence of recurrent ulcer bleeding in
patients with non–aspirin-related peptic ulcer bleeding
(12). Would the acid-suppressive effects of intravenous and
oral proton-pump inhibitors confer sufficient protection to
allow early resumption of aspirin therapy after successful
endoscopic hemostasis?
In this study, we hypothesized that after successful en-
doscopic control of ulcer bleeding, continuous aspirin ther-
apy would not be not inferior to stopping aspirin therapy
in terms of risk for recurrent bleeding in the presence of
proton-pump inhibitors.
METHODS
Design
We conducted a parallel randomized, placebo-controlled
study in which both patients and clinicians were blinded to
the treatment assignment. The clinical research ethics com-
mittee (institutional review board) of the Faculty of Med-
icine at the Chinese University of Hong Kong approved
the study protocol. All patients or their legal representa-
tives provided written, informed consent for participation
in the trial, and the study complied with the Declaration of
Helsinki, International Conference on Harmonisation of
Technical requirements for registration of pharmaceuticals
for human use Good Clinical Practice guidelines, and local
regulations. We recruited and followed patients from Feb-
ruary 2003 to October 2006. We vouch for the complete-
ness and veracity of the data and data analysis.
Setting and Participants
We conducted this single-center study at a university
medical center. Members of the gastroenterology team at the
Endoscopy Center of Prince of Wales Hospital (Sha Tin,
New Territories, Hong Kong) evaluated consecutive patients
who presented with overt signs of upper gastrointestinal bleed-
2 5 January 2010 Annals of Internal Medicine Volume 152 • Number 1
ing (melena or hematemesis) from February 2003 to Septem-
ber 2006 for inclusion in the trial. Patients were eligible if they
had a peptic ulcer showing active bleeding, visible blood ves-
sels, or adherent clots that were successfully treated by endo-
scopic therapy and continued to require low-dose aspirin
(ⱕ325 mg/d) for prophylaxis or treatment of cardiovascular
diseases. The indications for low-dose aspirin included pro-
phylaxis of established cardiovascular or cerebrovascular dis-
eases that required regular antiplatelet therapy. Patients who
received aspirin for primary prophylaxis were ineligible. We
excluded patients who had unsuccessful endoscopic hemosta-
sis of bleeding ulcers; those with gastric outlet obstruction,
ulcer perforation, known sensitivity to proton-pump inhibi-
tors, or previous partial gastrectomy or vagotomy; those re-
ceiving concomitant anticoagulant, corticosteroid, and non-
steroidal anti-inflammatory drugs; and those who were
pregnant. We did not exclude patients who received clopi-
dogrel in conjunction with aspirin, but we discontinued clo-
pidogrel therapy after randomization until the ulcer healed
completely. All inclusion and exclusion criteria were fulfilled
before patients were enrolled.
All patients received endoscopy within 24 hours of onset
of upper gastrointestinal bleeding. Endoscopic therapy in-
cluded epinephrine injection and thermal coagulation. We
injected aliquots of epinephrine, 0.5 mL to 1 mL (dilution,
1:10 000), around the bleeding vessel by using a 23-gauge
sclerotherapy needle until the bleeding stopped. We then ap-
plied thermal coagulation to vessels with a 3.2-mm heater
probe (model CD-10Z, Olympus Japan, Tokyo, Japan). If
bleeding had stopped and bleeding vessels were flattened or
cavitated, hemostasis was established. We removed clots cov-
ering ulcer craters and treated underlying vessels (if exposed).
We obtained biopsy specimens from the antrum and corpus
to determine whether H. pylori infection was present by using
rapid urease test and histologic findings. All patients received
an intravenous bolus injection of pantoprazole, 80 mg, fol-
lowed by infusion of pantoprazole, 8 mg/h, for 72 hours. We
followed this treatment with oral pantoprazole, 40 mg/d, until
the end of the study.
Randomization and Interventions
Randomization took place after endoscopic hemostasis
had been achieved. We randomly assigned eligible patients
to receive aspirin, 80 mg/d, or placebo for 8 weeks. We did
not stratify patients during randomization. An indepen-
dent committee generated the random allocation sequence
by using a computer-generated list of random numbers.
Number blocks were not used. To ensure concealed allo-
cation, an independent staff dispensed consecutively num-
bered, identically designed treatment packs that contained
sealed bottles of study drugs (aspirin or identical matching
placebo tablets). During hospitalization, a designated team
of physicians and surgeons who were unaware of treatment
assignment managed all study participants. We examined
patients who were clinically suspected to have developed
www.annals.org
 Continuation of Low-Dose Aspirin in Peptic Ulcer Bleeding
recurrent gastrointestinal bleeding by endoscopy. We did
not permit antiplatelet coprescription during follow-up.
Follow-up Procedures and Monitoring
After randomization, a designated team of physicians and
surgeons who were unaware of treatment assignment man-
aged the patients. During hospitalization, 1 investigator and 1
research nurse assessed the patients twice daily for any self-
reported adverse events. We assessed the intensity of adverse
events as mild (easily tolerated), moderate (interfered with
normal activities), or severe (incapacitating) and determined
the potential causality as unlikely, possible, or probable. A
serious adverse event was any event that threatened a patient’s
life, required prolonged hospitalization or rehospitalization, or
resulted in persistent disability or death. After discharge, we
provided 2 telephone hotlines for urgent inquires and report-
ing of adverse events. Patients returned for follow-up 30 and
56 days after discharge. We assessed drug compliance by pill
counts. We kept participants’ information anonymous and
identified participants by their study numbers and initials.
The study coordinator entered the data in the case report
forms into the electronic database within 48 hours. Only the
principal investigator and the study coordinator could access
the data, and only the principal investigator could make
changes to the electronic database. The database was locked
after independent verification and was backed up once every 2
weeks by the server in our center.
Outcomes
Our primary end point was recurrent peptic ulcer bleed-
ing within 30 days of endoscopic treatment. Patients received
another endoscopic examination if they had recurrent hemate-
mesis with vomiting of fresh blood; melena after a normal
stool; a decrease in hemoglobin level greater than 2 g/dL
within 24 hours, despite 2 or more units of blood transfused;
or unstable hemodynamic status (systolic blood pressure ⱕ90
mm Hg or pulse ⱖ110 beats/min) after achieving stabiliza-
tion. Recurrent ulcer bleeding was the presence of 1 or more
of these clinical features and confirmed by endoscopic evi-
dence, which included arterial spurter, nonbleeding visible
vessel, adherent clot, or fresh blood in the stomach.
Secondary end points occurring during the 8-week
study period included all-cause mortality; death attributed
to cardiovascular, cerebrovascular, or gastrointestinal com-
plications; requirement of blood transfusion; duration of
hospital stay (measured from day of recruitment); require-
ment of surgery; and recurrence of acute ischemic events
(the acute coronary syndrome and cerebrovascular acci-
dent). We diagnosed the acute coronary syndrome accord-
ing to the American College of Cardiology guidelines,
which included unstable angina, myocardial infarction
without ST-segment elevation, and myocardial infarction
with ST-segment elevation (13). We diagnosed cerebrovas-
cular accident according to the World Health Organiza-
tion criteria (14). An independent, blinded adjudication
committee ascertained whether recurrent ulcer bleeding
and atherothrombotic events had occurred according to
www.annals.org
Article
the prespecified criteria and verified the causality of deaths.
We included only events that were confirmed by an inde-
pendent, blinded adjudication committee in the analysis.
Statistical Analysis
We previously showed that among patients with peptic
ulcer bleeding who did not receive aspirin, 6.7% had recur-
rent ulcer bleeding within the 30 days after endoscopic ther-
apy followed by high-dose proton-pump inhibitor therapy
(12). Sample size estimation was based on the assumption that
6.7% of aspirin recipients who stopped aspirin therapy would
develop recurrent bleeding in 30 days and that continuous
aspirin therapy would not be inferior to stopping aspirin ther-
apy if the upper limit of the 95% CI of the difference in
recurrent bleeding did not exceed 10 percentage points. A
sample size of 75 patients per group would give the study a
power of 80% at a 5% level of significance with use of a
1-sided equivalence test of proportions (PASS software, ver-
sion 2008, NCSS, Kaysville, Utah). We allowed a wide non-
inferiority margin of 10 percentage points because recurrent
ulcer bleeding is potentially treatable, whereas interruption of
aspirin therapy may lead to more serious cardiovascular out-
comes. Our previous study showed that the 30-day recurrent
bleeding rate was up to 22.5% after endoscopic therapy with-
out high-dose proton-pump inhibitors (12). An independent
data safety and monitoring committee did 1 planned interim
analysis of the primary and secondary end points in Novem-
ber 2005 to compare the safety of the 2 treatments. If 1 treat-
ment was markedly inferior to the other (in terms of signifi-
cant increase in recurrent bleeding), we used a predefined
early stopping rule that specified termination of the trial if the
analysis reached a level of significance of 0.025. The result of
the interim analysis did not lead to protocol amendment or
early termination of the trial.
We used the Kaplan–Meier method to estimate the like-
lihood of reaching the end point of recurrent upper gastroin-
testinal bleeding within 30 days according to the intention-to-
treat population (all patients who had received at least 1 dose
of study medication). We also used the Kaplan–Meier
method to compare the 2 groups for all-cause mortality and
combined cardiovascular, cerebrovascular, and gastrointestinal
mortality within 8 weeks. Statistical tests for demographic
data and secondary end points included the t test, Mann–
Whitney U test, chi-square test, and Fisher exact test where
appropriate. We did all analyses by using SPSS, version 14
(SPSS, Hong Kong).
Role of the Funding Source
An independent educational grant from the Institute
of Digestive Disease, Chinese University of Hong Kong,
supported our study. Altana Pharma, Hong Kong, pro-
vided the pantoprazole that we used in our study. We
received no financial support from industry, and Altana
Pharma had no role in the design of the study, collec-
tion of data, statistical analysis, manuscript preparation
or interpretation, or decision to submit the manuscript
for publication.
5 January 2010 Annals of Internal Medicine Volume 152 • Number 1 3
Article Continuation of Low-Dose Aspirin in Peptic Ulcer Bleeding

RESULTS related bleeding events. We enrolled 156 patients in the

From February 2003 to September 2006, 3412 con- study: 78 in the aspirin group and 78 in the placebo group
secutive patients received a diagnosis of upper gastrointes- (Figure 1). Patients in each group did not receive crossover
tinal bleeding. Among them, 267 patients had aspirin- intervention during the trial. The 2 groups were similar

Figure 1. Study flow diagram.

Patients who presented with upper gastrointestinal

bleeding (from February 2003 to September 2006)
(n = 3412)

Confirmed peptic ulcer bleeding with high-risk

stigmata of recent hemorrhage
(n = 705)*

Peptic ulcer bleeding with stigmata and achieved

successful endoscopic hemostasis
(n = 664)

Nonaspirin or NSAID recipient

(n = 230)
NSAID or OTC NSAID
recipient (n = 167)
Bleeding due to aspirin-induced ulcer
(n = 267)

Randomly assigned Excluded patients (n = 111)

patients (n = 156) Allergic to aspirin: 1
Intestinal obstruction: 1
Needed double antiplatelet
therapy: 1
Ulcer perforation: 1
Aspirin Placebo Previous gastric surgery: 4
(n = 78) (n = 78) Concomitant use of
anticoagulant drugs: 7
Terminal cancer: 11
Suspected Suspected Moribund conditions: 34
clinical clinical Unable to or declined
recurrent recurrent consent: 45
bleeding bleeding Miscellaneous: 6
within 30 d within 30 d
Early termination:
(n = 13) (n = 9)
Needed double
antiplatelet
(n = 1) Early termination:
Needed warfarin Withdrew
(n = 1) consent
(n = 1)

Confirmed Confirmed
recurrent recurrent
bleeding bleeding
Died within 30 d Died within 30 d
(n = 1) (n = 8) (n = 10) (n = 4)

NSAID ⫽ nonsteroidal anti-inflammatory drug; OTC ⫽ over the counter.

* High-risk stigmata of hemorrhage included active spurting or oozing ulcer or ulcer showing protuberant vessel or adherent blood clot.
4 5 January 2010 Annals of Internal Medicine Volume 152 • Number 1 www.annals.org
 Continuation of Low-Dose Aspirin in Peptic Ulcer Bleeding Article

during hospitalization for other medical conditions (Table 1).

Table 1. Participant Characteristics
None of the patients we recruited received other antiplatelet
agents, such as clopidogrel, additional proton-pump inhibi-
Characteristic Aspirin Placebo
Recipients Recipients tors, or antacid therapy. We did not give nonsteroidal anti-
(n ⴝ 78) (n ⴝ 78) inflammatory drugs, warfarin, or corticosteroids to any patient
Men, n (%) 48 (62) 49 (63) throughout the study.
Mean age (SD), y 74 (9) 74 (8)
The adjudication committee evaluated 22 cases of sus-
Alcohol use, n (%) 6 (8) 6 (8)
Current smoker, n (%) 6 (8) 11 (14) pected recurrent upper gastrointestinal bleeding. The com-
American Society of Anesthesiologists grade mittee identified 12 cases of confirmed recurrent bleeding:
1 0 0
2 43 50
8 in the aspirin group and 4 in the placebo group. Among
3 34 26 patients with confirmed recurrent bleeding in the aspirin
4 1 2 group, 1 case was from a gastric ulcer and 7 were from
5 0 0
Indication for aspirin, n (%)
duodenal ulcers. All cases of confirmed recurrent bleeding
Cardiovascular diseases 40 (52) 47 (60) in the placebo group were from duodenal ulcers (Table 2).
Cerebrovascular diseases 30 (38) 23 (30) The 30-day cumulative incidence of recurrent ulcer bleed-
Both 8 (10) 8 (10)
Previous NSAID use, n (%)* 12 (15.4) 13 (16.7) ing in the intention-to-treat population was 10.3% (CI,
Helicobacter pylori positive, n (%) 31 (39.7) 33 (42.3) 3.4 to 17.2) in the aspirin group and 5.4% (CI, 0.3 to
Previous ulcer bleeding, n (%) 9 (12) 9 (12)
Mean baseline hemoglobin level (SD), g/dL 9.1 (2.4) 8.4 (2.2)
10.5) in the placebo group (difference, 4.9 percentage
Bled during hospital stay, n (%) 12 (15.3) 11 (14.1) points [CI, ⫺3.6 to 13.4 percentage points]; hazard ratio,
Location of bleeding ulcer, n (%) 1.9, [CI, 0.6 to 6.0]) (Figure 2). The site of recurrent
Gastric ulcer 43 (55) 41 (53)
Duodenal ulcer 34 (44) 35 (45)
bleeding was the same site as the original bleeding event.
Dieulafoy lesion 1 (1) 2 (3) Three patients (4.7%) with recurrent bleeding ulcers and
Endoscopic stigmata of bleeding, n (%) 61 patients (95.3%) who did not have recurrent bleeding
Active bleeding 24 (31) 27 (35)
Visible vessel 35 (45) 32 (41) tested positive for H. pylori.
Adherent clot 19 (24) 19 (24) Ten patients did not meet the prespecified criteria for
Mean size of ulcer (SD), cm 1.2 (0.8) 1.2 (0.6) recurrent upper gastrointestinal bleeding (5 patients in the
Ulcer ⱖ2 cm, n (%) 14 (18) 16 (21)
aspirin group and 5 in the placebo group). Among
NSAID ⫽ nonsteroidal anti-inflammatory drug. them, 8 patients had no endoscopic evidence of recur-
* Exposure to NSAID for ⬎1 wk and ⱕ1 y before presenting with upper gastro- rent bleeding and 2 (in the placebo group) did not have
intestinal bleeding.
endoscopy (1 patient had recurrent hematemesis and
with respect to demographic characteristics. More than died before arriving at the hospital, and 1 patient devel-
87% of patients had 90% or more drug compliance. About oped recurrent melena but was too ill to have further
15% of participants developed gastrointestinal bleeding endoscopic examination).

Table 2. Primary and Secondary End Points

End Point Aspirin Placebo Difference (95% CI)*

Recipients Recipients
(n ⴝ 78) (n ⴝ 78)
Suspected recurrent bleeding in 30 d, n (%) 13 (16.8) 9 (12.0) –
Confirmed recurrent bleeding in 30 d, n (%) 8 (10.3) 4 (5.4) 4.9 (⫺3.6 to 13.4)†
Gastric ulcer/duodenal ulcer, n/n 1/7 0/4 –
Stigmata of recent hemorrhage, n
Active bleeding 3 3 –
Visible vessel 2 1 –
Adherent clot 3 0 –
Median units of blood transfused (range), n 2 (0 to 10) 3 (0 to 9) 0 (⫺1.0 to 0.0)‡
Surgery, n (%) 0 (0) 1 (1.3) 1.3 (⫺6.5 to 12.1)
Median hospital stay (range), d 5 (3 to 25) 4.5 (1 to 45) 1 (0.0 to 1.0)†
Death, n (%)
30 d 1 (1.3) 7 (9) 11.6 (3.7 to 19.5)†
56 d 1 (1.3) 10 (12.9) 11.6 (3.7 to 19.5)†
Cause of death, n
Cardiovascular complications 1 5 –
Gastrointestinal complications 0 3 –
Pneumonia 0 2 –

* When the difference is between 2 percentages, it is expressed as percentage points.

† 95% CIs are Kaplan–Meier estimates.
‡ Difference in medians (95% CI of the difference).

www.annals.org 5 January 2010 Annals of Internal Medicine Volume 152 • Number 1 5

Article
Figure 2. Kaplan–Meier estimates of the incidence of
confirmed recurrent upper GI bleeding within 30 days.
0.5
Proportion of Recurrent Upper GI Bleeding
0.4
0.3
0.2
0.1
0.0
0 5
Patients at
risk, n
Aspirin 78
Placebo 78
Solid circles indicate censoring. GI ⫽ gastrointestinal.
The total number of units of blood transfused was
similar between the 2 treatment groups (Table 2). Only 1
patient in the placebo group required surgery because of a
perforated ulcer. The duration of hospital stay was also
similar between the 2 groups (Table 2).
One patient in the aspirin group died 30 days after
randomization. The patient was a woman aged 78 years
who had a history of ischemic heart disease and gangrene
of the toes. She developed an aspirin-related bleeding du-
odenal ulcer and died of congestive heart failure despite
successful endoscopic hemostasis. In the placebo group, 10
patients died (3 within 1 week, 4 between 1 week and 30
days, and 3 between 30 days and 8 weeks). These included
5 patients who died of vascular complications (2 died of
the acute coronary syndrome on days 1 and 7, 1 of recur-
rent stroke on day 12, and 2 of congestive heart failure on
days 20 and 39), 3 who died of gastrointestinal complica-
tions (2 died of perforated peptic ulcers on days 15 and 16
and 1 of uncontrolled bleeding on day 2), and 2 who died
of pneumonia (1 on day 35 and 1 on day 56). We did not
do co-intervention, such as percutaneous coronary angio-
plasty, before these patients died. The 30-day mortality
rate was lower in the aspirin group than in the placebo
group (1.3% [CI, 0% to 3.8%] vs. 9% [CI, 2.7% to
15.3%]; difference, 7.7 percentage points [CI, 0.9 to 14.5
percentage points]; hazard ratio, 0.2 [CI, 0.05 to 0.90]).
The Kaplan–Meier estimate of all-cause mortality at 8
weeks was lower in the aspirin group than the placebo
group (1.3% [CI, 0% to 3.8%] vs. 12.9% [CI, 5.5 to
20.3]; difference, 11.6 percentage points [CI, 3.7 to 19.5
6 5 January 2010 Annals of Internal Medicine Volume 152 • Number 1
Continuation of Low-Dose Aspirin in Peptic Ulcer Bleeding
percentage points]; hazard ratio, 0.2 [CI, 0.06 to 0.60])
(Figure 3, top). The mortality rate attributed to cardiovas-
cular, cerebrovascular, or gastrointestinal complications
was lower in the aspirin group than in the placebo group
(1.3% [CI, 0% to 3.8%] vs. 10.3% [CI, 3.4% to 17.2%];
Log-rank test (P = 0.25)
difference, 9 percentage points [CI, 1.7 to 16.3 percentage
Hazard ratio, 1.9 (95% CI, 0.6–6.0)
points]; hazard ratio, 0.2 [CI, 0.05 to 0.70]) (Figure 3,
bottom). In the primary analysis of confirmed recurrent
bleeding in 30 days, patients who died were censored at the
time of death if they had not experienced recurrent bleed-
Aspirin ing beforehand; a sensitivity analysis was done considering
Placebo death as a competing end point (Pepe and Mori test) (15).
The adjusted cumulative incidence of recurrent upper gas-
trointestinal bleeding at 30 days was 10.3% for patients
who received aspirin and 5.1% for patients who received
placebo (Pepe and Mori test P ⫽ 0.174) (15).
Six nonfatal, recurrent acute ischemic events were
10 15 20 25 30
reported (2 in the aspirin group and 4 in the placebo
Follow-up, d group): 3 patients had the acute coronary syndrome and
3 had acute stroke. A total of 14 patients in the aspirin
group had other adverse events (1 had a vasovagal at-
73 71 71 69 68 67 tack, 1 had type 2 respiratory failure, 1 had a seizure, 1
75 72 71 68 67 67
had gout, 2 had fever, 2 had dizziness, 1 had cough, 1
had chest infection, 1 had ankle edema, 1 had anemia,
and 2 had nausea and vomiting). Three patients in the
placebo group had adverse events (1 had a hallucination
and 2 had chest infection).
DISCUSSION
We show that, in the presence of proton-pump inhib-
itors and endoscopic therapy, continuing aspirin therapy
in patients with peptic ulcer bleeding was not equivalent
to stopping aspirin therapy in terms of risk for recurrent
ulcer bleeding. However, prolonged discontinuation of
aspirin therapy in these patients led to higher mortality
rates.
Aspirin and antiplatelet agents have been widely used
in the secondary prevention of acute coronary syndrome
and ischemic stroke, especially after interventional thera-
pies. Searching MEDLINE for studies published in
English for guideline recommendations until July 2009, the
American Heart Association/American College of Cardiol-
ogy and American Stroke Association updated guidelines
recommend starting or continuing aspirin therapy, 75 to
162 mg, indefinitely unless contraindicated for myocardial
infarction (16) and starting aspirin treatment at the initial
dose of 325 mg within 24 to 48 hours after onset of isch-
emic stroke (17). In patients with increased risk for bleed-
ing, the guidelines recommend using a lower dose of aspi-
rin but did not mention whether aspirin therapy should be
discontinued for some period. On the other hand, “a thor-
ough review of any medications with special attention to
the use of anticoagulants, antiplatelet agents, or medica-
tions associated with gastrointestinal hemorrhage should be
performed” (18).
www.annals.org
 Continuation of Low-Dose Aspirin in Peptic Ulcer Bleeding Article

In high-risk patients with cardiovascular diseases, the
timing of discontinuing antiplatelet treatment has often
been problematic. A recent study found that patients in
whom antiplatelet therapy was withheld before coronary
artery bypass graft surgery had a higher incidence of car-
diovascular complications than those who continued anti-
platelet therapy (19). However, perioperative administra-
tion of aspirin or other antithrombotic therapy has been
found to increase major bleeding complications (20, 21).
Although the risk for bleeding complications needs to bal-
ance with the risk for thrombosis, we did not find guide-
lines on whether aspirin therapy should be continued in
patients with peptic ulcer bleeding.
In our study, early resumption of aspirin therapy led
to a 50% increased risk for recurrent bleeding within 1
month (from 5.4% in the placebo group to 10.3% in the
aspirin group). Despite a higher risk for recurrent bleeding
from the peptic ulcer, only 1 of 78 patients who resumed
aspirin therapy early after endoscopic therapy died of
causes not related to bleeding. In contrast, discontinuing
mation by a study designed to address the optimal time
to restart antiplatelet therapy.
Our study has several limitations. First, the sample size
of 156 patients is relatively small. The patients we targeted
were older and often had several comorbid conditions, and
many were unable or unwilling to give informed consent
for the study. However, the 156 patients who we recruited
represented this critically ill group, as indicated by their age
and American Society of Anesthesiology grade. Although
our results suggest that continued aspirin therapy may in-
crease the risk for recurrent bleeding, prolonged discontin-
uation leads to adverse cardiovascular and cerebrovascular
outcomes, which are more serious. Second, the study de-
Figure 3. Kaplan–Meier estimates of the incidence of
mortality within 8 weeks.
0.5 Log-rank test (P = 0.005)
Hazard ratio, 0.2 (95% CI, 0.06–0.60)

Probability of All-Cause Mortality

aspirin therapy did not prevent ulcer mortality in 3 pa- 0.4

tients who received placebo therapy, despite use of a

proton-pump inhibitor. The small number of deaths 0.3
would restrict further interpretation of the results on mor- Aspirin
tality rates. Yet, the transfusion requirements between the Placebo
0.2
2 treatments were almost identical, which implies that re-
current bleeding was relatively mild and did not affect clin-
ical outcome of these patients. 0.1

In contrast, mortality rates were higher when aspirin

therapy was discontinued until ulcers healed. Most deaths 0.0
were related to cardiovascular events. We speculate that 0 7 14 21 28 35 42 49 56
gastrointestinal bleeding might lead to a higher mortality Follow-up, d
Patients at
rate in the placebo group because these patients were more risk, n
vulnerable to atherothrombosis and therefore could not Aspirin 78 77 77 77 76 75 75 75 75
tolerate bleeding well. The fact that none of the patients Placebo 78 75 74 70 70 69 68 68 67
who received aspirin and only 3 who received placebo Log-rank test (P = 0.016)
0.5
died of gastrointestinal complications could be a chance
Probability of Death Related to CVS, CVA,

Hazard ratio, 0.2 (95% CI, 0.05–0.70)

finding. The difference in mortality rates, however, re-
mained significant even if we excluded death due to 0.4

gastrointestinal complications. The protective effect of

or GI Conditions

antiplatelet agents seems to outweigh their potential gas- 0.3

trointestinal toxicity. Aspirin

Unlike most cases of death related to gastrointestinal Placebo

0.2
bleeding that occurred usually within the first 3 to 5 days
after index bleeding (9), death in the placebo group of
0.1
happened throughout 8 weeks of follow-up. This phenom-
enon may be related to the fact that, despite rapid clearance
of aspirin from the circulation, the antiplatelet effects of 0.0

aspirin last for at least a few days because of the perma- 0 7 14 21 28 35 42 49 56

Follow-up, d
nent inactivation of the platelets’ cyclooxygenase activ- Patients at
ity on prostaglandin synthase 1 and synthase 2 (11). risk, n
Aspirin 78 77 77 77 76 75 75 75 75
One might infer that aspirin can be discontinued for 3
Placebo 78 75 74 70 70 69 68 68 67
to 5 days after index bleeding and resumed after stabi-
lization. This strategy, which in theory minimizes the Solid circles indicate censoring. CVA ⫽ cerebrovascular; CVS ⫽ cardio-
bleeding risk and vascular ischemia risk, needs confir- vascular; GI ⫽ gastrointestinal.
www.annals.org 5 January 2010 Annals of Internal Medicine Volume 152 • Number 1 7
Article Continuation of Low-Dose Aspirin in Peptic Ulcer Bleeding
sign did not allow us to compare the protective effects of
proton-pump inhibitors with aspirin because both treat-
ment groups received high-dose intravenous pantoprazole
followed by oral pantoprazole. Given the published efficacy
of proton-pump inhibitors, it would be unethical to study
the effects of early resumption of aspirin in the absence of
an effective protective agent. Yet, the low recurrent bleed-
ing rate in both treatment groups compared with previous
studies suggests that there were substantial protective ef-
fects (12). Third, we used low-dose aspirin (80 mg);
whether our results can be extrapolated to a higher dose of
165 mg to 320 mg is uncertain. However, available evi-
dence supports the use of a daily dose of aspirin in the
range of 75 to 100 mg for the long-term prevention of
serious vascular events in high-risk patients (16, 17). Pru-
dent use of low-dose aspirin in patients with history of
peptic ulcer bleeding is warranted. Finally, 2 patients in the
placebo group had symptoms of recurrent bleeding but
were not included in the primary analyses as having recur-
rent bleeding because they did not have endoscopy. Add-
ing these 2 cases as recurrent bleeding will further reduce
the difference between the 2 groups.
In conclusion, among patients with peptic ulcer bleed-
ing who received low-dose aspirin, continuous aspirin ther-
apy may increase the risk for recurrent bleeding. However,
antiplatelet agents potentially reduce overall mortality.
Early resumption of low-dose aspirin therapy with proton-
pump inhibitors in patients with bleeding ulcers and car-
diovascular diseases should be considered.
From the Institute of Digestive Disease, Chinese University of Hong
Kong, Sha Tin, New Territories, Hong Kong.
Grant Support: By an independent educational grant from the Institute
of Digestive Disease, Chinese University of Hong Kong. Altana Pharma,
Hong Kong, provided pantoprazole.
Potential Conflicts of Interest: Consultancies: F.K.L. Chan (Pfizer, Ot-
suka). Honoraria: F.K.L. Chan (Pfizer, Takeda, AstraZeneca). Grants
pending: F.K.L. Chan (Takeda). Patents pending: J.J.Y. Sung (Nycomed).
Other: F.K.L. Chan (chairman of the steering committee for Condor).
Reproducible Research Statement: Study protocol: Available to ap-
proved individuals through written agreements with the Institute of Di-
gestive Disease, Chinese University of Hong Kong. Statistical code: Avail-
able from Dr. Sung (e-mail, joesung@cuhk.edu.hk). Data set: Not available.
Requests for Single Reprints: Joseph J.Y. Sung, MD, PhD, Institute of
Digestive Disease, Prince of Wales Hospital, New Territories, Hong
Kong; e-mail, joesung@cuhk.edu.hk.
Current author addresses and author contributions are available at www
.annals.org.
References
1. Weil J, Colin-Jones D, Langman M, Lawson D, Logan R, Murphy M, et al.
Prophylactic aspirin and risk of peptic ulcer bleeding. BMJ. 1995;310:827-30.
[PMID: 7711618]
8 5 January 2010 Annals of Internal Medicine Volume 152 • Number 1
2. Hallas J, Dall M, Andries A, Andersen BS, Aalykke C, Hansen JM, et al. Use
of single and combined antithrombotic therapy and risk of serious upper gastro-
intestinal bleeding: population based case-control study. BMJ. 2006;333:726.
[PMID: 16984924]
3. Laine L. Review article: gastrointestinal bleeding with low-dose aspirin—
what’s the risk? Aliment Pharmacol Ther. 2006;24:897-908. [PMID: 16948802]
4. Kelly JP, Kaufman DW, Jurgelon JM, Sheehan J, Koff RS, Shapiro S. Risk
of aspirin-associated major upper-gastrointestinal bleeding with enteric-coated or
buffered product. Lancet. 1996;348:1413-6. [PMID: 8937281]
5. Derry S, Loke YK. Risk of gastrointestinal haemorrhage with long term use of
aspirin: meta-analysis. BMJ. 2000;321:1183-7. [PMID: 11073508]
6. Yeomans ND, Lanas AI, Talley NJ, Thomson AB, Daneshjoo R, Eriksson B,
et al. Prevalence and incidence of gastroduodenal ulcers during treatment with
vascular protective doses of aspirin. Aliment Pharmacol Ther. 2005;22:795-801.
[PMID: 16225488]
7. Chan FK, Chung SC, Suen BY, Lee YT, Leung WK, Leung VK, et al.
Preventing recurrent upper gastrointestinal bleeding in patients with Helicobacter
pylori infection who are taking low-dose aspirin or naproxen. N Engl J Med.
2001;344:967-73. [PMID: 11274623]
8. Lanas A, Garcı́a-Rodrı́guez LA, Arroyo MT, Gomollón F, Feu F, González-
Pérez A, et al; Asociación Española de Gastroenterologı́a. Risk of upper gastro-
intestinal ulcer bleeding associated with selective cyclo-oxygenase-2 inhibitors,
traditional non-aspirin non-steroidal anti-inflammatory drugs, aspirin and com-
binations. Gut. 2006;55:1731-8. [PMID: 16687434]
9. Lanas A, Garcı́a-Rodrı́guez LA, Arroyo MT, Bujanda L, Gomollón F, Forné
M, et al; Investigators of the Asociación Española de Gastroenterologı́a (AEG).
Effect of antisecretory drugs and nitrates on the risk of ulcer bleeding associated
with nonsteroidal anti-inflammatory drugs, antiplatelet agents, and anticoagu-
lants. Am J Gastroenterol. 2007;102:507-15. [PMID: 17338735]
10. Ng FH, Wong SY, Lam KF, Chang CM, Lau YK, Chu WM, et al. Gas-
trointestinal bleeding in patients receiving a combination of aspirin, clopidogrel,
and enoxaparin in acute coronary syndrome. Am J Gastroenterol. 2008;103:865-
71. [PMID: 18177451]
11. Patrono C, Garcı́a Rodrı́guez LA, Landolfi R, Baigent C. Low-dose aspirin
for the prevention of atherothrombosis. N Engl J Med. 2005;353:2373-83.
[PMID: 16319386]
12. Lau JY, Sung JJ, Lee KK, Yung MY, Wong SK, Wu JC, et al. Effect of
intravenous omeprazole on recurrent bleeding after endoscopic treatment of
bleeding peptic ulcers. N Engl J Med. 2000;343:310-6. [PMID: 10922420]
13. Cannon CP, Battler A, Brindis RG, Cox JL, Ellis SG, Every NR, et al.
American College of Cardiology key data elements and definitions for measuring
the clinical management and outcomes of patients with acute coronary syn-
dromes. A report of the American College of Cardiology Task Force on Clinical
Data Standards (Acute Coronary Syndromes Writing Committee). J Am Coll
Cardiol. 2001;38:2114-30. [PMID: 11738323]
14. Hatano S. Experience from a multicentre stroke register: a preliminary report.
Bull World Health Organ. 1976;54:541-53. [PMID: 1088404]
15. Pepe MS, Mori M. Kaplan-Meier, marginal or conditional probability curves
in summarizing competing risks failure time data? Stat Med. 1993;12:737-51.
[PMID: 8516591]
16. Antman EM, Hand M, Armstrong PW, Bates ER, Green LA, Halasyamani
LK, et al; 2004 Writing Committee Members. 2007 Focused Update of the
ACC/AHA 2004 Guidelines for the Management of Patients With ST-Elevation
Myocardial Infarction: a report of the American College of Cardiology/American
Heart Association Task Force on Practice Guidelines: developed in collaboration
With the Canadian Cardiovascular Society endorsed by the American Academy
of Family Physicians: 2007 Writing Group to Review New Evidence and Update
the ACC/AHA 2004 Guidelines for the Management of Patients With ST-
Elevation Myocardial Infarction, Writing on Behalf of the 2004 Writing Com-
mittee. Circulation. 2008;117:296-329. [PMID: 18071078]
17. Adams HP Jr, del Zoppo G, Alberts MJ, Bhatt DL, Brass L, Furlan A, et al;
American Heart Association/American Stroke Association Stroke Council.
Guidelines for the early management of adults with ischemic stroke: a guideline
from the American Heart Association/American Stroke Association Stroke Coun-
cil, Clinical Cardiology Council, Cardiovascular Radiology and Intervention
Council, and the Atherosclerotic Peripheral Vascular Disease and Quality of Care
Outcomes in Research Interdisciplinary Working Groups: The American Acad-
emy of Neurology affirms the value of this guideline as an educational tool for
neurologists. Circulation. 2007;115:e478-534. [PMID: 17515473]
18. Adler DG, Leighton JA, Davila RE, Hirota WK, Jacobson BC, Qureshi
www.annals.org
 Continuation of Low-Dose Aspirin in Peptic Ulcer Bleeding Article
WA, et al; ASGE. ASGE guideline: The role of endoscopy in acute non-variceal 20. Ghaffarinejad MH, Fazelifar AF, Shirvani SM, Asdaghpoor E, Fazeli F,
upper-GI hemorrhage. Gastrointest Endosc. 2004;60:497-504. [PMID: Bonakdar HR, et al. The effect of preoperative aspirin use on postoperative
15472669] bleeding and perioperative myocardial infarction in patients undergoing coronary
19. Leiva-Pons JL, Carrillo-Calvillo J, Leiva-Garza JL, Loyo-Olivo MA, artery bypass surgery. Cardiol J. 2007;14:453-7. [PMID: 18651504]
Piãa-Ramı́rez BM, López-Quijano JM, et al. [Importance of the time for 21. Manzano-Fernández S, Pastor FJ, Marı́n F, Cambronero F, Caro C,
stopping the combined use of aspirin and clopidogrel in patients under- Pascual-Figal DA, et al. Increased major bleeding complications related to triple
going coronary artery by-pass graft surgery]. Arch Cardiol Mex. 2008;78: antithrombotic therapy usage in patients with atrial fibrillation undergoing percu-
178-86. [PMID: 18754409] taneous coronary artery stenting. Chest. 2008;134:559-67. [PMID: 18641090]

www.annals.org 5 January 2010 Annals of Internal Medicine Volume 152 • Number 1 9

Annals of Internal Medicine
Current Author Addresses: Drs. Sung, Lau, Wu, Chiu, Wong, and
Chan and Ms. Ching: Institute of Digestive Disease, Prince of Wales
Hospital, New Territories, Hong Kong.
Dr. Lee: Endoscopy Center, Baptist Hospital, 222 Waterloo Road,
Kowloon, Hong Kong.
Dr. Leung: Department of Medicine, United Christian Hospital,
Kowloon, Hong Kong.
Author Contributions: Conception and design: J.J.Y. Sung, J.Y.W. Lau,
J.C.Y. Wu, Y.T. Lee, F.K.L. Chan.
Analysis and interpretation of the data: J.J.Y. Sung, J.Y.L. Ching, J.C.Y.
Wu, F.K.L. Chan.
Drafting of the article: J.J.Y. Sung, J.C.Y. Wu, P.W.Y. Chiu, F.K.L.
Chan.
Critical revision of the article for important intellectual content: J.J.Y.
Sung, J.Y.W. Lau, J.C.Y. Wu, F.K.L. Chan.
Final approval of the article: J.J.Y. Sung, J.Y.W. Lau, J.C.Y. Wu, V.K.S.
Leung, F.K.L. Chan.
Provision of study materials or patients: J.J.Y. Sung, J.C.Y. Wu, V.K.S.
Leung, F.K.L. Chan.
Administrative, technical, or logistic support: J.J.Y. Sung, J.Y.L. Ching,
J.C.Y. Wu, V.K.S. Leung, F.K.L. Chan.
Collection and assembly of data: J.J.Y. Sung, J.Y.L. Ching, J.C.Y. Wu,
Y.T. Lee, P.W.Y. Chiu, V.K.S. Leung, V.W.S. Wong.

www.annals.org 5 January 2010 Annals of Internal Medicine Volume 152 • Number 1 W-1