’Original article

Randomized clinical trial: the impact of

gastrointestinal risk factor screening and
prophylactic proton pump inhibitor therapy in
patients receiving dual antiplatelet therapy
Berit E.S. Jensena, Jane M. Hansena, Kasper S. Larsenc, Anders B. Junkerb, Jens F. Lassend, Svend E. Jensene
and Ove B. Schaffalitzky de Muckadella

Objective Dual antiplatelet therapy reduces the risk of ischemic complications after acute coronary syndrome, but increases the
risk of bleeding including upper gastrointestinal bleeding (UGIB). The aim of this study was to examine the effect of screening for
risk of UGIB and prophylactic proton pump inhibitor (PPI) treatment in dual-antiplatelet-treated patients at risk of UGIB and to
assess the significance of dual antiplatelet therapy compliance for cardiovascular events.
Patients and methods In a register-based randomized-controlled trial, 2009 patients were included at the time of first
percutaneous coronary intervention and randomized to either screening or control. Screened high-risk patients were prescribed
pantoprazole 40 mg during the 1-year after percutaneous coronary intervention.
Results The incidence of UGIB was 0.8 versus 1.3% in screened patients and controls, respectively (P = 0.381). Significantly
fewer screened patients (5.4%) than controls (8.0%) underwent upper gastrointestinal endoscopy (P = 0.026). Screened patients
(2.9%) had significantly fewer events of unstable angina pectoris than controls (4.7%) (P = 0.036) and a higher compliance to dual
antiplatelet therapy (88.3 vs. 85.0%) (P = 0.035), but no statistically difference was observed in the incidences of myocardial
infarction and all-cause mortality (1.0 vs. 1.5%) (P = 0.422).
Conclusion Screening for risk factors for UGIB and subsequent prophylactic PPI treatment did not significantly reduce the
incidence of UGIB. Prescription of PPI was associated with a higher compliance with dual antiplatelet therapy and decreases the
risk of recurrent cardiovascular events. Eur J Gastroenterol Hepatol 00:000–000
Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.

Introduction Recently, 65% of first-time PCI and dual-antiplatelet-

Upper gastrointestinal bleeding (UGIB) is a severe condi-
tion, with almost 2000 hospital admissions each year in
Denmark (population: 5.61 million) and a 30-day mor-
tality of 10% [1]. Dual antiplatelet therapy with ADP-
receptor inhibitor and low-dose aspirin reduces the risk of
ischemic complications after acute coronary syndrome [2]
and is the standard treatment after percutaneous coronary
intervention (PCI) and stenting [3]. However, dual anti-
platelet therapy increases the risk of gastrointestinal (GI)
bleeding [4,5]. Up to 43% of all ulcer-bleeding admissions
are related to antithrombotic treatment [1].
European Journal of Gastroenterology & Hepatology 2017, 00:000–000
Keywords: dual antiplatelet therapy, proton pump inhibitor,
upper gastrointestinal bleeding
Departments of aMedical Gastroenterology, bCardiology, Odense University
Hospital, cDepartment of Public Health, University of Southern Denmark, Odense,
d
Department of Cardiology, Aarhus University Hospital, Skejby and eDepartment of
Cardiology, Aalborg University Hospital, Aalborg, Denmark
Correspondence to Berit E.S. Jensen, MD, Department of Medical
Gastroenterology, Odense University Hospital, Sdr. Boulevard 29,
DK-5000 Odense C, Denmark
Tel: + 45 654 12755; fax: + 45 661 11328; e-mail: berit.elin@dadlnet.dk
Received 5 March 2017 Accepted 6 June 2017
Supplemental digital content is available for this article. Direct URL citations appear
in the printed text and are provided in the HTML and PDF versions of this article on
the journal's website (www.eurojgh.com).
treated patients were assessed to be at high risk of devel-
oping UGIB [6]. The risk factors for UGIB include a
history of peptic ulcer, older age, NSAIDs, selective ser-
otonin reuptake inhibitors, oral anticoagulants, and cor-
ticosteroids. These risk factors are well documented in
low-dose aspirin-treated patients [7,8].
Major bleeding reduces short-term [9] and long-term
survival after acute coronary syndrome [10], possible
because of termination of antiplatelet treatment after a
bleeding episode. Compliance with the antiplatelet treat-
ments is important to reduce the risk of further thrombotic
events [11] and proton pump inhibitor (PPI) may improve
compliance with antiplatelet treatment as PPI treatment
reduces dyspepsia related to low-dose aspirin treatment
[12]. About 20% of patients admitted with acute coronary
syndrome are subsequently treated with a PPI [13].
No studies have examined the effect of a systematic
screening for UGIB risk factors before dual antiplatelet
therapy. A register-based, randomized-controlled trial
allows for complete follow-up of the included patients
while still reflecting clinical practice [14]. The aims of this
study were to investigate: (a) the effect of a systematic risk
factor screening for ulcer bleeding and subsequent PPI
prophylaxis on the risk of UGIB among first-time PCI
patients who commence a 1-year dual antiplatelet treat-
ment; (b) whether PPI prophylaxis in high-risk patients can

0954-691X Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved. DOI: 10.1097/MEG.0000000000000934 1

Copyright r 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
2 European Journal of Gastroenterology & Hepatology
increase compliance with dual antiplatelet treatment and
reduce the risk of cardiovascular events.
Patients and methods
This was a multicenter register-based, randomized-
controlled trial of first-time PCI patients in the western
part of Denmark (population: 3.3 million) during 2-year
period from April 2011 to May 2013.
Patients
PCI-naive patients were included from all the primary PCI
centers in western Denmark. All patients were prescribed
1-year dual antiplatelet therapy with low-dose aspirin and
ADP-receptor inhibitor.
Exclusion criteria were as follows: (a) previous PCI;
(b) previous ADP-receptor inhibitor treatment (≥1 month
of ADP-receptor inhibitor treatment was allowed as most
patients initiate treatment before PCI); or (c) lack of
informed consent. Monotherapy with ADP-receptor inhi-
bitor because of intolerability of aspirin was accepted to
avoid selection bias.
Intervention and randomization
After PCI, potential participants received oral and written
study information. If the patient accepted participation, he
or she was subsequently contacted by phone by the prin-
cipal investigator for an explanation of detailed study
information and inclusion. The patients completed a
questionnaire to assess the risk factors for UGIB. To
increase the participation rate, nonresponders were sent a
reminder after 1 week. Using opaque envelopes, the ran-
domization was performed on the basis of a computer-
generated list. Patients were block-randomized according
to intervention center and 1 : 1 to screening or control
group. Central randomization and screening was chosen to
avoid affecting the treating physicians’ habits of prescrib-
ing PPI. Further, the treating cardiologist and the general
practitioner were not involved in any aspects of this study.
Questionnaire
With a few modifications, we used previously validated
questions from the HEP-FYN study [15] and the ques-
tionnaire was tested by face validity [16]. The ques-
tionnaire included questions on dyspepsia, ulcer history,
and drug use (Supplementary Appendix 1, Supplemental
digital content 1, http://links.lww.com/EJGH/A207).
Risk score
For screened patients, an assessment of risk of UGIB
(Table 1) was performed. Each risk factor was weighted on
the basis of previous studies [7,8] and assigned an indivi-
dual score. On the basis of the individual scores, a total
risk score was calculated. A high risk of UGIB was defined
as a total score of 2 or more.
Proton pump inhibitor prophylaxis
All screened patients at high risk of UGIB received a pre-
scription for pantoprazole 40 mg covering the entire study
period, along with thorough written information, by mail.
Copyright r 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Month 2017 • Volume 00 • Number 00
Table 1. Assessment of risk of ulcer bleeding
Score
Age
< 60 0
60–69 1
70–79 2
≥ 80
Dyspepsiaa
3
1
Previous uncomplicated ulcerb 2
Previous ulcer bleeding 3
NSAIDc 2
Corticosteroid 2
SSRI 2
Oral anticoagulant 2
SSRI, selective serotonin reuptake inhibitor.
a
Dyspepsia: pain in the upper part of the stomach, acid reflux, heartburn, or nausea/
vomiting.
b
Uncomplicated ulcer covers peptic ulcer without bleeding and is in this context
based on patient information.
c
Weekly or daily use.
Screened patients already treated with PPI were switched
to pantoprazole 40 mg (if possible) as pantoprazole is less
CYP2C19-inhibiting and thereby leads to reduced inter-
actions with the ADP-receptor inhibitor [17]. In the con-
trol group, PPI was allowed at the discretion of the treating
cardiologist or the general practitioner.
Follow-up
All patients were followed for 1 year from the date of PCI.
Using the unique personal identification number assigned
to all residents in Denmark, linking between the different
registers was possible [18]. The following registers were
used: (a) the Danish National Patient Register [19] holds
information on all inpatient and outpatient discharge
diagnoses. The diagnoses are encoded according to the
WHO [20] defined International Classification of Diseases,
10th revision (ICD-10) criteria. (b) The Danish Civil
Registration System [21] maintains records on dates of
birth, death, migrations, and current residence of all
Danish citizens. The Danish Civil Registration System was
used to assess all-cause mortality. (c) The Danish National
Database of Reimbursed Prescriptions [22], which is a
prescription database holding information on redeemed,
reimbursed prescriptions. The medications and quantities
are registered according to the WHO Anatomical
Therapeutic Chemical (ATC) system and the defined daily
doses [23]. (d) The Western Denmark Heart Registry [24]
is a clinical register containing detailed patient-specific and
procedure-specific information on all coronary interven-
tions performed in western Denmark. The Western
Denmark Heart Registry was also used to describe eligible
patients not recruited for inclusion to ensure the external
validity of the study results.
Outcome and definitions
The primary outcome was UGIB defined as hematemesis
or melena, history of hematemesis, or melena and low
hemoglobin levels or need for blood transfusion. The
secondary outcomes were uncomplicated ulcer (diagnosed
at endoscopy or operation with a diameter no <5 mm and
with loss of tissue), acute coronary syndrome defined
according to the ‘Academic Research Consortium’ [25],
 Screening and prophylaxis for ulcer bleeding Jensen et al.
and upper GI endoscopy. For detailed information on the
ICD-10 codes used, see Supplementary Table 1
(Supplemental digital content 2, http://links.lww.com/
EJGH/A208).
All GI events were validated by scrutinizing medical
records blinded to the randomization code.
Treatment and compliance with PPI (ATC A02BC) and
dual antiplatelet therapy (ATC B01AC04, B01AC06,
B01AC22, and B01AC24) were defined from the day of
the redeemed prescription and onwards, assuming inges-
tion of one defined daily dose a day. To allow for minor
noncompliance, a patient was considered to follow the
treatment with either PPI or dual antiplatelet therapy if he
or she redeemed more than 80% of the prescribed amount.
Delays in redeeming clopidogrel prescriptions are found in
up to one out of five patients [26]. To account for such a
redeeming pattern, this study allowed for a 14-day delay
after PCI.
Statistical analyses
The results were reported as absolute number and per-
centage. Estimated glomerular filtration rate, blood pres-
sure, and age were compared using two-sample t-tests after
testing for a normal distribution. Categorical variables
were compared using χ2-test and Fisher’s exact test when
relevant. Missing values was handled with median value
imputation for continuous variables. Numbers needed to
treat with screening to prevent one case of unstable angina
pectoris were calculated. A P-value of less than 0.05 was
considered statistically significant.
We carried out a sensitivity analysis as a peer protocol
analysis among the compliant PPI users to quantify how
the compliance with PPI treatment affected the outcomes
among patients at high risk.
Sample size calculation
Prophylaxis with PPI was expected to reduce the risk of
ulcer bleeding by two-thirds [12]. This population inclu-
ded patients already treated with PPI (20%). Assuming a
bleeding risk of 3% per year [4,5], it was estimated that
screening and subsequent PPI treatment of risk patients
could reduce the risk to 1.1% per year. The required
number of patients was 975 per group if 2α = 0.05 and
β = 0.80. Expected 1-year mortality after PCI was 5% [25].
In total, inclusion of 2000 patients was planned.
Ethics
All patients included signed an informed consent form.
Generally, PPI is well tolerated and leads to very few
adverse reactions [27,28].
The control group could receive treatment as usual
(i.e. they could receive PPI treatment at the general practi-
tioners’ or at other physicians’ discretion). The Regional
Committee of Health Ethics approved the study.
Furthermore, permission to establish a private research
register as well as permission to use register data was pro-
vided by the Danish Data Protection Agency. The study was
registered at ClinicalTrials.gov (number NCT01447498).
All authors had access to the study data and approved the
final manuscript.
Copyright r 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
www.eurojgh.com 3
Results
During 25 months of inclusion, 2013 patients were
enrolled. Four patients who experienced a GI event after
PCI but before randomization were excluded. A total of
2009 patients were included in the analyses. Figure 1
shows the study flow. The participation rate among invited
patients was 75%. Baseline characteristics are summarized
in Table 2.
Completeness of recruitment
The patients included were compared with all eligible
patients. The included group was characterized by a sig-
nificantly higher proportion of men, a higher BMI, and a
higher incidence of previous heart disease, but a younger
mean age, lower blood pressure, fewer smokers, and a
lower use of cholesterol-lowering drugs (Supplementary
Table 2, Supplemental digital content 3, http://links.lww.
com/EJGH/A209).
Effect of screening and proton pump inhibitor prophylaxis
in the prevention of upper gastrointestinal bleeding
There was a nonstatistically significant tendency for fewer
GI events among the screened patients compared with the
controls (0.8 vs. 1.3%) (Table 3). Statistically significantly
fewer patients in the screened group underwent upper
GI endoscopy during the follow-up period (Table 3).
Detailed information on drug use and risk profile for
patients with UGIB is presented in Supplementary Table 3
(Supplemental digital content 4, http://links.lww.com/
EJGH/A210).
Cardiovascular events and all-cause mortality
Screened patients had statistically significantly fewer
events of unstable angina pectoris. To prevent one case of
unstable angina pectoris, 5.6 patients had to be screened.
No differences were found with respect to myocardial
infarctions.
Comparison of screened patients and controls accord-
ing to all-cause mortality showed a tendency toward
higher mortality in the control group (10/997 vs.
15/1012). All deaths were among patients at high risk.
Incidences of UGIB and cardiovascular events among the
deceased patients were 0/25 and 2/25, respectively.
Compliance
In the screened group, 72.3% of patients complied with
PPI treatment (on average 340 defined daily doses over the
year), whereas almost all (94%) redeemed at least one
prescription for PPI (Table 4). Forty percent of controls at
high risk redeemed at least one prescription for PPI during
the follow-up period, and on average, they redeemed 103
defined daily doses over the year. Low-risk patients
redeemed less PPI than the high-risk patients and no dif-
ferences were found between screened and control groups.
Compliance with low-dose aspirin was similar in both
groups irrespective of randomization status or risk
assessment (50.7–55.8%) (Table 5). Compliance to ADP-
receptor inhibitor was higher in the screened group com-
pared with the control group: 865 (88.3%) versus 851
(85.0%) (P = 0.035) (Table 5).
4 European Journal of Gastroenterology & Hepatology Month 2017 • Volume 00 • Number 00

Fig. 1. Flow diagram of inclusion. DAPT, dual antiplatelet theraphy; PCI, percutaneous coronary intervention; WDHR, Western Denmark Heart Registry.

Sensitivity analysis
A peer protocol analysis among the compliant PPI users
(Supplementary Table 4, Supplemental digital content 5,
http://links.lww.com/EJGH/A211) showed no statistically
significant difference between the groups according to GI
events and all-cause mortality, but a clear tendency toward
more GI events was found in the control group.
Compliance with ADP-receptor inhibitor treatment was
significantly higher in the screened group compared with
the control group. The level of cardiovascular events did
not differ between the groups.
Discussion
This is the first prospective, randomized study testing the
effect of a systematic screening for risk factors for UGIB
and prophylactic prescription of PPI to PCI-naive patients
in dual antiplatelet therapy.

Copyright r 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
 Screening and prophylaxis for ulcer bleeding Jensen et al. www.eurojgh.com 5

This study uncovered a tendency toward a lower inci- may reflect a reduction of dyspepsia during PPI prophy-
dence of UGIB and uncomplicated ulcer in the screened laxis [12]. Patients in the screened group were more
group compared with the control group, but this did not compliant with ADP-receptor inhibitor treatment and had
reach statistical significance. This result is supported by the fewer unstable angina pectoris events.
results of a recently published case–control study [29]. This study has several strengths. First, this is a multi-
Significantly fewer patients in the screened group center, register-based, randomized, controlled trial based
underwent upper GI endoscopy during follow-up, which on a population covering more than half of all PCIs in
Denmark (58%). Second, the national public health
Table 2. Characteristics of included patients (2009 patients) authorities provide almost all healthcare-related contacts
Screened [n (%)] Controls [n (%)] in Denmark and all first-time PCIs in western Denmark
were available. This approach ensures external validity of
All 997 (100.0) 1012 (100.0)
Men 729 (73.1) 758 (74.9) the general applicability of the results. The register-based
Age [mean (SD) (range)] (years) 64.7 (10.2) (31–92) 64.8 (10.6) (34–89) approach ensured complete follow-up and very few miss-
eGFR [mean (SD) (range)] (ml/ 81.5 (35.5) (6–821) 79.7 (18.5) (5–192) ing data. The participation rate was reasonable and all age
min)
Smoking status groups were represented.
Current 253 (25.4) 279 (27.6) This study was not blinded or placebo controlled,
Previous 406 (40.7) 398 (39.3) which is a limitation. Placebo treatment could potentially
Never 297 (29.8) 293 (29.0)
Treatment with have resulted in fewer visits at general practitioners and
Antidiabetics 113 (11.3) 103 (10.2) thereby to lower PPI use in the control group. However,
Cholesterol-lowering drugs 502 (50.4) 496 (49.0) we do not believe that this has had a significant influence
Antihypertensive drugs 535 (53.7) 546 (54.0)
ADP-I on our results as the patients in the control group were not
Clopidogrel 559 (56.1) 571 (56.4) informed about their randomization status or risk
Ticagrelor 434 (43.5) 439 (43.4) assessment.
Prasugrel 4 (0.4) 2 (0.2)
PPI (before PCI) 189 (19.0) 182 (18.0)
The event rate was considerably lower (1.3%) than esti-
Aspirin (before PCI) 470 (47.1) 450 (44.5) mated (3.0%). Validation of GI outcomes from medical
History of heart disease records has confirmed the positive findings, but no investi-
Heart surgery 31 (3.1) 43 (4.2)
Acute coronary syndrome 57 (5.7) 62 (6.1)
gation was made of the frequency or reasons for false-
BMI negative diagnoses because of which the number of GI out-
< 25 279 (28.0) 279 (27.6) comes could be under-reported. However, a wide range of GI
25–30 425 (42.6) 421 (41.6)
> 30 221 (22.2) 248 (24.5)
diagnoses were extracted and validated (Supplementary
Blood pressure Table 1, Supplemental digital content 2, http://links.lww.
Systolic [mean (SD) (range)] 138.1 (21.6) 139.6 (21.5) com/EJGH/A208); therefore, we do not believe that many
(60–220) (60–230)
Diastolic [mean (SD) (range)] 77.7 (12.3) 77.7 (12.0) (40–120)
outcomes were missed. In our study, validation of ulcer
(34–117) diagnoses from Danish National Patient Register by scruti-
History ofb nizing medical records showed incorrect diagnoses in 30% of
Gastrointestinal symptoms 578 (58.0) 591 (58.4)
Peptic ulcer 108 (10.8) 98 (9.7)
all cases, of which 44% resulted in a positive UGIB diag-
Ulcer bleeding 28 (2.8) 25 (2.5) nosis. This is in accordance with the validation from a pre-
Current use ofb vious study [30]. Data from the entire population of first-time
NSAID 130 (13.0) 170 (16.8)
SSRI 57 (5.7) 46 (4.5)
PCI patients in Western Denmark Heart Registry confirm the
Corticosteroid 28 (2.8) 33 (3.3) incidence of UGIB found in this study (1%), although this
Oral anticoagulant 49 (4.9) 71 (7.0) was not validated. The number of patients using PPI in the
Increased riskc 672 (67.4) 710 (70.2)
Days from PCI to randomization 11 (0–94) 11 (2–58)
control group was higher than expected, and could have
contributed toward the low UGIB rate. Altogether, the lower
ADP-I, ADP-receptor inhibitor; eGFR, estimated glomerular filtration rate; than anticipated event rate resulted in an underpowered
PCI, percutaneous coronary intervention; PPI, proton pump inhibitor; SSRI,
selective serotonin reuptake inhibitor.
study, which may explain the lack of significant differences in
b
According to the questionnaire. the UGIB incidence.
c
Patients with a total score 2 or more are considered to be at increased risk of Not all eligible patients were included. The patients
developing upper gastrointestinal bleeding. included deviated slightly from eligible patients, but the

Table 3. Outcomes
Screened [n/N (%) (95% CI)] Controls [n/N (%) (95% CI)] P-value
Gastrointestinal eventsa
Upper GI bleeding 8/997 (0.8) (0.3–1.6) 13/1012 (1.3) (0.7–2.2) 0.381
Uncomplicated ulcer 2/997 (0.2) (0.2–0.7) 4/1012 (0.4) (0.1–1.0) 0.687
Upper GI endoscopyb 54/997 (5.4) (4.1–7.0) 81/1012 (8.0) (6.4–9.9) 0.026
Cardiovascular events
Unstable angina pectoris 29/997 (2.9) (2.0–4.2) 48/1012 (4.7) (3.5–6.2) 0.036
Myocardial infarction 133/997 (13.3) (11.3–15.6) 146/1012 (14.4) (12.3–16.7) 0.519
All-cause mortality 10/997 (1.0) (0.5–1.8) 15/1012 (1.5) (0.8–2.4) 0.422

CI, confidence interval; GI, gastrointestinal.

a
No ulcer perforations were observed.
b
During the follow-up period.
P < 0.05 was considered statistically significant.

Copyright r 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
6 European Journal of Gastroenterology & Hepatology Month 2017 • Volume 00 • Number 00

Table 4. Compliance to proton pump inhibitor

Screened [n (%)] Controls [n (%)]

Low risk Increased risk Low risk Increased risk

All 325 (100.0) 672 (100.0) 302 (100.0) 710 (100.0)
PPI any time during follow-up 55 (16.9) 632 (94.0) 48 (15.9) 288 (40.6)
PPI no later than 14 days after inclusion 19 (5.8) 542 (80.7) 6 (2.0) 53 (7.5)
Compliant 80% of follow-upa 18 (5.5) 486 (72.3) 15 (5.0) 121 (17.0)
Cumulated DDD during follow-up per patient 35 340 34 103
Treated with PPI before inclusion 22 (6.8) 167 (24.9) 23 (7.6) 159 (22.4)

DDD, defined daily doses; PPI, proton pump inhibitor.

a
More than 80% of the follow-up period.

Table 5. Compliance to low-dose aspirin and adenosine-diphosphate-receptor inhibitor

Screened Controls

Low risk Increased risk Low risk Increased risk

All ADP-I users 320 (100.0) 660 (100.0) 298 (100.0) 703 (100.0)
Compliant ADP-I usersa,b 291 (90.9) 574 (87.0) 260(87.2) 591 (84.1)
Cumulated DDD ADP-I users (per patient) 347 339 340 332
All aspirin users 325 (100) 672 (100) 302 (100) 710 (100)
Compliant aspirin usersa 181 (55.7) 375 (55.8) 153 (50.7) 399 (56.2)
Cumulated DDD aspirin users (per patient) 272 265 260 262

ADP-I, ADP-receptor inhibitor; DDD, defined daily doses.

a
More than 80% of the follow-up period.
b
ADP-I compliance was significantly different between those screened and controls (P = 0.035), and between those screened and controls at increased risk (P = 0.027).

differences observed were small and statistical significance

was probably caused by the large sample size. By con-
sidering the individual variables, the included patients are
believed to be a representative sample for the population.
Also, the PCI indications (24% STEMI, 27% NSTEMI,
47% stable angina, and 2% others) reflect the indications
for the general population [31]. However, the incidence of
all-cause mortality in the population included was low
(1.2%) compared with more than 5% in comparable
populations [25,31] and that of the entire population of
first-time PCI patients in the Western Denmark Heart
Registry (5.8%). Therefore, patients included in this study
appear to be healthier than patients in the general popu-
lation. This could have caused fewer GI events compared
with the entire population, resulting in loss of power, and
could also compromise the external validity. If all eligible
patients had been enrolled in this study, we would have
expected a higher rate of cardiovascular events and death.
Testing PCI patients for Helicobacter pylori infection in
this study to ensure eradication before PPI prescription
was not possible, but it would have been reasonable
according to the suggestions from the Maastricht con-
sensus panel. However, this study should reflect real life, in
which patients should commence prophylactic PPI as soon
as possible. As a consequence of a low H. pylori prevalence
in Denmark (< 20% [15]) and only a possible effect of PPI
on H. pylori-positive duodenal ulcers [32], the effect of
H. pylori testing and treatment with PPI in this study was
expected to be minimal.
Over-the-counter sales of PPI in Denmark amount to
only 3% of the total sales [33] and are not considered to
influence the results of this study.
Although only 72.3% (672 patients) of the screened
patients at high risk prescribed prophylactic PPI followed
the treatment regime (redeemed > 80% of the prescri-
bed amount), patients in the compliant group redeemed
340 defined daily doses during the follow-up (covering on
average 93% of the days). Noncompliance with PPI
treatment is well known among patients treated with PPI
for asymptomatic reflux disease [34]. The absence of GI
symptoms is likely contributing toward noncompliance
with PPI treatment in this study as many patients had no
GI symptoms at baseline. The cost of the PPI could also
have contributed toward nonredemption of prescriptions,
but the individual copayment is small because of reim-
bursement from the Danish health authorities. More con-
trols than expected redeemed PPI during follow-up;
however, only 17% redeemed more than 80% of the
prescribed amount and the average amount redeemed was
only 103 defined daily doses during follow-up among
high-risk controls.
A peer protocol analysis of PPI-compliant patients did
not alter our results, with the exception of cases of
unstable angina pectoris, which is no longer statistically
lower among PPI-compliant patients. We believe that this
is because of loss of power. The percentage of UGIB
among PPI-compliant patients in this analysis is higher
than that in the main analysis. Possible explanations
include confounding by indication, but the total number of
events is too small (n = 15) to make general assumptions.
Compliance with low-dose aspirin treatment was low.
Patients treated with low-dose aspirin before PCI may not
redeem the prescription before current stock is used, but
this was accounted for in this study. Use of drugs pre-
scribed for family members, or over-the-counter use, could
not be accounted for, which could explain an apparently
irregular usage patterns for low-dose aspirin. GI symptoms
are known to affect compliance with low-dose aspirin
treatment [35]; however, compliance with low-dose
aspirin measured as redeemed drugs was not altered by
the intervention in this study.

Copyright r 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
 Screening and prophylaxis for ulcer bleeding Jensen et al.
Screening and PPI prophylaxis was associated with a
higher compliance with treatment involving ADP-receptor
inhibitor. This finding is probably explained by a reduced
tendency to develop antiplatelet related dyspepsia among
PPI-treated patients.
Information on drug use under hospital admissions was
not available, thus reducing apparent compliance during
the follow-up period. However, there is no reason to
believe that this would impact the results as this is believed
to affect both groups equally.
Conclusion
Screening for risk factors for UGIB and subsequent pro-
phylactic PPI treatment did not reduce the incidence of
UGIB. However, some of the secondary outcomes (endo-
scopy and unstable angina pectoris) achieved statistically
significant reductions in the screened group. To prevent
one case of unstable angina pectoris, 5.6 patients needed to
be screened. With this in mind and considering the scarcity
of PPI adverse reactions to PPI, that UGIB is a serious
condition, and that PPI prophylaxis was associated with
increased compliance with ADP-receptor inhibitor treat-
ment and fewer cardiovascular events, the administration
of PPI prophylaxis to patients at high risk of UGIB seems
reasonable.
Acknowledgements
Berit E.S. Jensen is the main author of the manuscript and
has been in charge of data collection, statistical analyses,
and interpretation of data; Jane M. Hansen, Berit E.S.
Jensen, Anders B. Junker, Jens F. Lassen, Svend E. Jensen,
and Ove B. Schaffalitzky de Muckadell developed the
study concept and design and were all involved in critical
revision of the manuscript for important intellectual con-
tent and interpretation of data; Kasper S. Larsen was
involved in statistical analyses, critical revision of the
manuscript for important intellectual content, and inter-
pretation of data. All authors have approved the
final draft.
This study was funded by the Danish Research Council,
the Danish Heart Association, the Region of Southern
Denmark, and the Odense University Hospital Research
Council. None of the above-mentioned institutions were
involved in the study design, collection, analyses, or
interpretation of data.
Conflicts of interest
Kasper S. Larsen has served as an advisory board member
for Menarini. Anders B. Junker has served as a speaker
and an advisory board member for AstraZeneca. Jens F.
Lassen has served as a speaker and an advisory board
member for AstraZeneca. Ove B. Schaffalitzky de
Muckadell has served as an advisor for Novo Nordisk. For
the remaining authors there are no conflicts of interest.
References
1 Danish Clinical Register of Emergency Surgery (DCRES). Annual report
of Danish Clinical Register of Emergency Surgery (DCRES). DCRES;
Copyright r 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
www.eurojgh.com 7
2013. Available at: https://www.sundhed.dk/sundhedsfaglig/kvalitet/
kliniskekvalitetsdatabaser/akutte-sygdomme/mavesaar-nip/.
2 Peters RJG, Mehta SR, Fox KAA, Zhao F, Lewis BS, Kopecky SL, et al.
Effects of aspirin dose when used alone or in combination with clopi-
dogrel in patients with acute coronary syndromes. Circulation 2003;
108:1682–1687.
3 The Danish Society of Cardiology. Available at: http://nbv.cardio.dk/aks.
4 Yusuf S, Zhao F, Metha SR, Chrolavicius S, Tognoni G, Fox KK. Effects
of clopidogrel in addition to aspirin in patients with acute coronary
syndromes without ST-segment elevation. N Engl J Med 2001;
345:494–502.
5 Mortensen J, Thygesen SS, Johnsen SP, Vinther PM, Kristensen SD,
Refsgaard J. Incidence of bleeding in ‘real-life’ acute coronary syndrome
patients treated with antithrombotic therapy. Cardiology 2008;
111:41–46.
6 Søltoft Jensen BE, Hansen JM, Søltoft Larsen K, Junker AB, Lassen JF,
Jensen SE, et al. High prevalence of risk factors of ulcer bleeding in dual
antiplatelet treated patients after percutaneous coronary intervention.
Dan Med J 2015; 62:A5092.
7 Rodríguez LAG, Hernández-Díaz S, De Abajo FJ. Association between
aspirin and upper gastrointestinal complications: systematic review of
epidemiologic studies. Br J Clin Pharmacol 2001; 52:563–571.
8 Gargallo CJ, Sostres C, Lanas A. Prevention and treatment of NSAID
gastropathy. Curr Treat Options Gastroenterol 2014; 12:398–413.
9 Moscucci M, Fox KAA, Cannon CP, Klein W, López-Sendón J,
Montalescot G, et al. Predictors of major bleeding in acute coronary
syndromes: the Global Registry of Acute Coronary Events (GRACE). Eur
Heart J 2003; 24:1815–1823.
10 Segev A, Strauss BH, Tan M, Constance C, Langer A, Goodman SG.
Predictors and 1-year outcome of major bleeding in patients with non-
ST-elevation acute coronary syndromes: insights from the Canadian
Acute Coronary Syndrome Registries. Am Heart J 2005; 150:690–694.
11 Biondi-Zoccai GGL, Lotrionte M, Agostoni P, Abbate A, Fusaro M,
Burzotta F, et al. A systematic review and meta-analysis on the hazards
of discontinuing or not adhering to aspirin among 50,279 patients at risk
for coronary artery disease. Eur Heart J 2006; 27:2667–2674.
12 Yeomans N, Lanas A, Labenz J, van Zanten SV, van Rensburg C,
Racz I, et al. Efficacy of esomeprazole (20 mg once daily) for reducing
the risk of gastroduodenal ulcers associated with continuous use of
low-dose aspirin. Am J Gastroenterol 2008; 103:2465–2473.
13 Schreiner GC, Laine L, Murphy S, Cannon C. Evaluation of proton pump
inhibitor use in patients with acute coronary syndromes based on risk
factors for gastrointestinal bleed. Crit Pathw Cardiol 2009; 6:169–172.
14 Thuesen L, Jensen LO, Tilsted HH, Maeng M, Terkelsen C, Thayssen P,
et al. Event detection using population-based health care databases in
randomized clinical trials: a novel research tool in interventional cardi-
ology. Clin Epidemiol 2013; 5:357–361.
15 Wildner-Christensen M, Møller Hansen J, Schaffalitzky De Muckadell OB.
Rates of dyspepsia one year after Helicobacter pylori screening and
eradication in a Danish population. Gastroenterology 2003; 125:
372–379.
16 De Vet HCW, Terwee CB, Mokkink LB, Knol DL. Measurement in
medicine: a practical guide. Cambridge, NY: Cambridge University
Press; 2011.
17 Drepper MD, Spahr L, Frossard JL. Clopidogrel and proton pump
inhibitors – where do we stand in 2012? World J Gastroenterol 2012;
18:2161–2171.
18 Thygesen LC, Ersbøll AK. Danish population-based registers for public
health and health-related welfare research: introduction to the supple-
ment. Scand J Public Health 2011; 39 (Suppl): 8–10.
19 Lynge E, Sandegaard JL, Rebolj M. The Danish National Patient
Register. Scand J Public Health 2011; 39 (Suppl): 30–33.
20 World Health Organization. International classification of diseases 10th
version (ICD-10). WHO; Available at: http://www.who.int/classifications/
icd/en/.
21 Pedersen CB. The Danish Civil Registration System. Scand J Public
Health 2011; 39 (Suppl): 22–25.
22 Johannesdottir SA, Horvath-Puho E, Ehrenstein V, Schmidt M,
Pedersen L, Sorensen HT. Existing data sources for clinical epide-
miology: the Danish National Database of Reimbursed Prescriptions.
Clin Epidemiol 2012; 4:303–313.
23 WHO Collaborating Centre for Drug Statistics Methodology. ATC/DDD
index 2014 and guidelines for ATC classification and DDD assignment.
Oslo: Norwegian Institute of Public Health. Available at: http://www.
whocc.no/atc_ddd_publications/guidelines/.
8 European Journal of Gastroenterology & Hepatology
24 Schmidt M, Maeng M, Jakobsen CJ, Madsen M, Thuesen L,
Nielsen PH, et al. Existing data sources for clinical epidemiology: the
Western Denmark Heart Registry. Clin Epidemiol 2010; 2:137–144.
25 Jensen LO, Maeng M, Kaltoft A, Thayssen P, Hansen HHT, Bottcher M,
et al. Stent thrombosis, myocardial infarction, and death after drug-
eluting and bare-metal stent coronary interventions. J Am Coll Cardiol
2007; 50:463–470.
26 Roth GA, Morden NE, Zhou W, Malenka DJ, Skinner J. Clopidogrel use
and early outcomes among older patients receiving a drug-eluting
coronary artery stent. Circ Cardiovasc Qual Outcomes 2012;
5:103–112.
27 Vanderhoff BT, Tahboub RM. Proton pump inhibitors: an update. Am
Fam Physician 2002; 66:273–280.
28 Karlstadt RG, Walker K. Update on prescribing information for panto-
prazole. Am Fam Physician 2003; 67:1189–1190. Author reply 1190.
29 Jiang Z, Wu H, Duan Z, Wang Z, Hu K, Ye F, et al. Proton-pump
inhibitors can decrease gastrointestinal bleeding after percutaneous
coronary intervention. Clin Res Hepatol Gastroenterol 2013;
37:636–641.
30 Lassen A, Hallas J, Schaffalitzky, de Muckadell OB. Complicated
and uncomplicated peptic ulcers in a Danish county 1993–2002:
Copyright r 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Month 2017 • Volume 00 • Number 00
a population-based cohort study. Am J Gastroenterol 2006; 101:
945–953.
31 VestDansk Hjertedatabase. Årsrapport 2013 VestDansk Hjertedatabase
[in Danish]. Western Denmark Heart Registry; 2013. Available at: http:
//vdhd.dk/wp-content/uploads/2014/10/%C385rsrapport2013.pdf.
[Accessed 1 May 2015].
32 Sostres C, Carrera-Lasfuentes P, Benito R, Roncales P, Arruebo M,
Arroyo MT, et al. Peptic ulcer bleeding risk. The role of Helicobacter
Pylori infection in NSAID/low-dose aspirin users. Am J Gastroenterol
2015; 110:684–689.
33 Haastrup P, Paulsen MS, Zwisler JE, Begtrup LM, Hansen JM,
Rasmussen S, et al. Rapidly increasing prescribing of proton pump
inhibitors in primary care despite interventions: a nationwide
observational study. Euro J Gen Pract 2014; 20:290–293.
34 Domingues G, Moraes-Filho JP. Noncompliance is an impact factor in
the treatment of gastroesophageal reflux disease. Expert Rev
Gastroenterol Hepatol 2014; 8:761–765.
35 Pratt S, Thompson VJ, Elkin EP, Naesdal J, Sorstadius E. The impact of
upper gastrointestinal symptoms on nonadherence to, and dis-
continuation of, low-dose acetylsalicylic acid in patients with
cardiovascular risk. Am J Cardiovasc Drugs 2010; 10:281–288.