THE LANCET
Articles
Randomised trial of eradication of Helicobacter pylori
before non-steroidal anti-inflammatory drug therapy to prevent
peptic ulcers
Francis K L Chan, Joseph J Y Sung, S C Sydney Chung, K F To, M Y Yung, Vincent K S Leung, Y T Lee,
Cynthia S Y Chan, Edmund K M Li, Jean Woo
Summary
Background Helicobacter pylori infection is common in
patients with peptic ulcers caused by the use of non-
steroidal anti-inflammatory drugs (NSAIDs). But the
pathogenic role of H pylori in this disease is controversial.
We studied the efficacy of eradication of H pylori in the
prevention of NSAID-induced peptic ulcers.
Methods We recruited patients with musculoskeletal pain
who required NSAID treatment. None of the patients had
previous exposure to NSAID therapy. Patients who had
H pylori infection but no pre-existing ulcers on endoscopy
were randomly allocated naproxen alone (750 mg daily) for
8 weeks or a 1-week course of triple therapy (bismuth
subcitrate 120 mg, tetracycline 500 mg, metronidazole
400 mg, each given orally four times daily) before
administration of naproxen (750 mg daily). Endoscopy was
repeated after 8 weeks of naproxen treatment or when
naproxen treatment was stopped early because of bleeding
or intractable dyspepsia. All endoscopic examinations were
done by one endoscopist who was unaware of treatment
assignment. The primary endpoint was the cumulative rate
of gastric and duodenal ulcers.
Findings 202 patients underwent endoscopic screening for
enrolment in the trial, and 100 eligible patients were
randomly assigned treatment. 92 patients completed the
trial (47 in the naproxen group, 45 in the triple-therapy
group). At 8 weeks, H pylori had been eradicated from no
patients in the naproxen group and 40 (89%) in the triple-
therapy group (p<0·001). 12 (26%) naproxen-group
patients developed ulcers: five had ulcer pain and one
developed ulcer bleeding. Only three (7%) patients on
triple therapy had ulcers, and two of these patients had
failure of H pylori eradication (p=0·01). Thus, 12 (26%)
patients with persistent H pylori infection but only one
(3%) with successful H pylori eradication developed ulcers
with naproxen (p=0·002).
Interpretation Eradication of H pylori before NSAID therapy
reduces the occurrence of NSAID-induced peptic ulcers.
Lancet 1997; 350: 975–79
Departments of Medicine (F K L Chan MRCP, J J Y Sung MD,
V K S Leung MRCP, Y T Lee MRCP, E K M Li FRCP, J Woo MD), Surgery
(S C S Chung MD, M Y Yung BN), and Anatomical and Cellular
Pathology (K F To FRCPA), Prince of Wales Hospital, Shatin, New
Territories, Hong Kong; and Community and Family Medicine, The
Chinese University of Hong Kong, Hong Kong (C S Y Chan MD)
Correspondence to: Dr Francis K L Chan
Vol 350 • October 4, 1997
Introduction
Peptic ulcer disease caused by non-steroidal anti-
inflammatory drug (NSAID) therapy is a health-care
issue worldwide. In the UK, about 1·5 million people
aged over 60 years take NSAIDs at any one time.1
Patients who take NSAIDs have a four-fold to six-fold
increased risk of developing peptic ulcers.2,3 Every year,
about 12 000 ulcer complications occur in the UK as a
result of NSAID therapy.4 Although chronic NSAID
users are at increased risk of ulcer disease, those who take
intermittent short-course NSAID therapy are also at risk
of the disease. Most ulcer complications tend to develop
in the first few weeks of therapy, particularly among
individuals without previous exposure to NSAIDs.3,5 The
risk of ulcer complications is greatest among the elderly
and patients with comorbid disease.2,3 Up to 60% of ulcer
complications occur without antecedent symptoms.6 The
annual direct medical costs associated with these
complications is about $3·9 billion in the USA.7
Previous studies have shown that prophylactic therapy
with misoprostol, famotidine, and omeprazole is effective
in the prevention of NSAID-induced ulcers.8–10 However,
these prophylactic regimens are very expensive.11 Cost-
effective measures to prevent NSAID-induced ulcer
disease are not available.
H pylori is present in about 50% of patients with
NSAID-associated ulcer disease.12,13 Studies of the
interaction between H pylori infection and NSAIDs have
reported conflicting findings.12–26 Whether H pylori
infection increases the risk of ulcer development in users
of NSAIDs is not known.
The aim of our study was to investigate whether
eradication of H pylori before the start of NSAID therapy
reduced the occurrence of gastroduodenal ulcers in
patients without previous exposure to NSAIDs. We
postulated that underlying infection with H pylori
increases the predisposition of NSAID users to develop
ulcers. Thus, eradication of H pylori could protect these
patients from developing peptic ulcers.
Methods
Patients with musculoskeletal disorders that required NSAID
therapy were eligible for the study. We recruited patients from
the medical outpatient clinic at Prince of Wales Hospital, Hong
Kong, and from the family clinic of the Chinese University of
Hong Kong. We excluded patients if they: were younger than 18
years; had been previously exposed to NSAIDs (including
aspirin) for longer than 1 month; had taken NSAIDs (including
aspirin), antiulcer drugs, steroid, anticoagulants, or cytotoxic
agents in the previous 8 weeks; had received antihelicobacter
therapy; had a history of peptic ulcer disease or gastric surgery;
or if they had renal impairment (serum creatinine >200 ␮mol/L).
Eligible patients were invited to take part in the study and were
975
THE LANCET
202 eligible patients underwent
endoscopic screening
93 excluded
H pylori negative (91)
Ulcers (2)
109 positive for
H pylori
9 refusals after
first endoscopy
100 patients randomised
50 naproxen 50 naproxen plus
alone triple therapy
5 default follow-up
3 default follow-up Did not need naproxen (2)
Refusal (2) Refusal (1)
Intolerance to Intolerance to naproxen (1)
naproxen (1) Intolerance to triple
therapy (1)
47 patients 45 patients
47 H pylori 40 H pylori
positive negative
12 with peptic ulcers 1 with
peptic ulcers peptic ulcers
Trial profile
given a detailed explanation. Those who gave their informed
written consent to take part in the study underwent upper-
gastroinestinal endoscopy. The study protocol was approved by
the ethics committee of the Chinese University of Hong Kong.
All endoscopic assessment was done by a one endoscopist
(FKLC) to eliminate between-observer variation. We defined a
peptic ulcer as a circumscribed mucosal break 5 mm or more in
diameter with a well-defined ulcer crater, whereas smaller or
superficial lesions were classified as erosions. Ulcer size was
measured by standard Olympus biopsy forceps (Olympus FG-
25K), with the fully open instrument equivalent to 5 mm. We
excluded patients with ulcers detected at baseline endoscopy.
Five random biopsy specimens were taken from the antrum;
two specimens were used for the rapid urease test (CLO test,
Delta West, Bentley, Western Australia) and three samples were
sent for histological assessment with haematoxylin and eosin
stain and Warthin-Starry stain. The histological assessment was
done by one pathologist (KFT) who was unaware of treatment
assignment. A patient was judged to be infected with H pylori if
they had a positive rapid urease test that was confirmed by the
histology. A positive rapid urease test alone was not deemed
sufficient for the diagnosis of H pylori infection.
Patients who were confirmed to have H pylori infection and
did not have an ulcer at baseline endoscopy were randomly
assigned to one of two treatment groups. We used a list of
computer-generated random numbers for treatment assignment.
Patients were assigned naproxen 750 mg daily in three divided
doses at 8 h intervals for 8 weeks, or a 1-week course of triple
therapy (bismuth subcitrate 120 mg, tetracycline 500 mg, and
metronidazole 400 mg, each given orally four times daily)
followed by naproxen 750 mg daily in three divided doses at 8 h
976
Naproxen alone Triple therapy plus p
(n=47) naproxen (n=45)
Men/women 15/32 11/34 0·43*
Median (range) age in years 61 (40–84) 64 (33–79) 0·50†
Current smokers 14 (30%) 12 (27%) 0·74*
Alcohol use 2 (4%) 2 (4%) 1·00‡
Osteoarthritis/rheumatoid 36/2/9 37/2/6 0·51*
arthritis/other
Comorbid illness§ 27 (57%) 31 (69%) 0·26*
Stomach
Subepithelial haemorrhage 0 1 1·00‡
Erosions 2 (4%) 7 (16%) 0·09‡
Duodenum
Subepithelial haemorrhage 0 0 1·00‡
Erosions 1 1 1·00‡
*Pearson ␹2 test. †Mann-Whitney U test. ‡Fisher’s exact test. §Comorbid illness
includes cerebrovascular accident, hypertension, ischaemic heart disease, heart
failure, diabetes mellitus, chronic obstructive airway disease, and liver cirrhosis.
11 (23%) patients in the naproxen group and 15 (33%) in the triple-therapy group
had more than one disease. No patients took aspirin or antacids before the study.
Table 1: Baseline characteristics of patients
intervals for 8 weeks. Dologesic (propoxyphene napsylate 50 mg,
paracetamol 325 mg) was given to patients for temporary
pain relief during triple therapy, or as an adjuvant therapy to
naproxen if requested by patients.
A research nurse interviewed each patient about their smoking
habits, alcohol consumption, drug history, and concurrent
medical illness. Complete blood count and renal function tests
were carried out. We assessed drug compliance by counting
unused tablets. Endoscopy was repeated by the same
endoscopist after 8 weeks of naproxen therapy to document
5 H pylori gastroduodenal ulcers. H pylori status was reassessed by
positive repeating antral biopsies. We defined eradication of H pylori
as negative rapid urease test and absence of the bacteria on
histology. The endoscopist and pathologist were unaware of the
treatment allocation and previous endoscopic or histological
2 with findings. Those patients who developed intractable dyspeptic
symptoms that required the early termination of naproxen, or
who had gastrointestinal haemorrhage (defined as haematemesis
or melaena) underwent endoscopy before 8 weeks.
The primary endpoint was the cumulative rate of gastric and
duodenal ulcers. All patients who returned for follow-up
endoscopy were included in the intention-to-treat analysis. We
did a per-protocol analysis for those patients who took more than
60% of the prescribed naproxen, or at least 5 days of the triple
therapy in patients on antihelicobacter treatment. We also
compared the clinical outcome of patients with or without
successful eradication of H pylori infection.
We estimated the sample size based on our previous study,
which showed an ulcer prevalence of 33% in NSAID users who
were infected with H pylori compared with 6% in NSAID users
who did not have H pylori infection.19 To achieve a statistical
power of 85%, with an ␣ error of 5%, 45 patients were needed in
each treatment group. Under the assumption of a drop-out rate
of 10% in each group, we calculated that 100 patients were
required in the study. The results of the two treatment groups
were compared by two-tailed Fisher’s exact test, Pearson ␹2 test,
and Mann-Whitney U test. A p value less than 0·05 was taken
as significant.
Results
202 consecutive patients agreed to take part and
underwent endoscopy for enrolment in the trial. Of these
patients, 91 were not infected by H pylori, two had ulcers
at initial endoscopy, and nine patients with H pylori
refused to participate in the study after endoscopy. 100
patients were randomly assigned treatment in two groups
of 50. Eight withdrew after randomisation and defaulted
follow-up (naproxen group: one refused to undergo
second endoscopy, one refused to take part, and one
Vol 350 • October 4, 1997

Naproxen alone Triple therapy plus
(n=47) naproxen (n=45)
Total number of ulcers 12 (26%) 3 (7%)
Gastric 9 3
Duodenal 2 0
Gastric and duodenal 1 0
Symptomatic ulcers 6 (13%) 1 (2%)
Pain 5 1‡
Bleeding 1 0
Stomach
Subepithelial haemorrhage 6 2
Erosions 15 18
Duodenum
Subepithelial haemorrhage 1 0
Erosions 4 6
*Pearson ␹2 test. †Fisher’s exact test. ‡Failure to eradicate H pylori infection.
Table 2: Intention-to-treat analysis
patient had abdominal pain after taking naproxen; triple
therapy plus naproxen group: adequate pain relief with
dologesic in two, one refused to take part, one patient
with palpitation after taking naproxen, and one patient
with intolerance to triple therapy). 92 patients completed
the study: 47 in the naproxen group and 45 in the triple-
therapy group.
The flow of patients is shown in the trial profile. The
two groups were well matched for age, sex, smoking and
drinking habits, and underlying arthritis (table 1). In the
triple-therapy group, there were more patients with
comorbid illness and gastric erosions were more frequent
on baseline endoscopy than in the naproxen group, but
the difference was not significant. 41 patients completed
at least 5 days of the prescribed 1-week triple therapy.
40 (89%) triple-therapy patients had successful
eradication of H pylori, versus none in the naproxen
group (p<0·001).
The cumulative rate of gastric and duodenal ulcers
after 8 weeks of treatment is shown in table 2. In the
intention-to-treat analysis, 12 (26%) patients in the
naproxen group compared with three (7%) in the triple-
therapy group developed peptic ulcers (p=0·01). The size
of ulcers ranged from 5 mm to 10 mm (mean diameter of
6 mm). Seven patients had more than one ulcer; one
patient had both gastric and duodenal ulcers. In the
naproxen group, six of the 12 patients developed
symptomatic ulcers that required early termination of
naproxen: five had intractable dyspepsia (two developed
ulcers after 2 weeks of naproxen, three had ulcers at
5–7 weeks), and one had ulcer bleeding during week 7.
Of the three patients who developed ulcers in the triple-
therapy group, two had failure of H pylori eradication
(one completed triple therapy but developed intractable
dyspepsia and an ulcer was detected after 1 week of
naproxen, and the other patient took triple therapy for
only 2 days because of gastrointestinal upset and an ulcer
was found on follow-up endoscopy at 8 weeks). Even
when we assumed that all defaulters (except those who
responded adequately to dologesic alone) were treatment
failures, pretreatment with antihelicobacter therapy still
significantly reduced the rate of ulcers: 15 patients with
ulcers in the naproxen group versus six in the triple-
therapy group (p=0·048). One patient in each group had
epigastric pain before 8 weeks, but endoscopy did not
reveal any ulcers. Other endoscopic lesions were also
detected (subepithelial haemorrhage and erosions), but
there was no significant between-group differences.
Comparison of the rate of ulcers in patients with and
without successful eradication of H pylori, showed that
Vol 350 • October 4, 1997
THE LANCET
p Naproxen alone Triple therapy plus p
(n=43) naproxen (n=38)
0·01* Total number of ulcers 12 (28%) 2 (5%) 0·007*
Gastric 9 2 0·04*
Duodenal 2 0
Gastric and duodenal 1 0
0·11† Symptomatic ulcers 6 (14%) 1 (3%)† 0·11‡
Pain 5 1
Bleeding 1 0
*Pearson ␹2 test. †Failure to eradicate H pylori infection. ‡Fisher’s exact test.
0·27† Table 3: Per-protocol analysis
0·55*
only one (2·5%) patient developed ulcer after successful
1·00†
0·52† eradication, whereas 12 (26%) patients infected with
H pylori developed ulcers with naproxen (p=0·002). Of
the 15 patients who developed peptic ulcers in both
treatment groups, 11 (73%) were older than 60 years
(range 49–84 years). Comorbid disease was present in 11
(73%) patients (seven had more than one disease).
The per-protocol analysis included 43 naproxen-group
patients and 38 triple-therapy patients (table 3). Four
patients in the naproxen group took less than 60% of
the drug prescribed: one had adequate pain relief with
dologesic alone, one did not have adequate pain relief
with naproxen, and two had dyspeptic symptoms that
disappeared when the frequency of naproxen intake was
reduced. In the triple-therapy group, three patients took
the antibiotics for less than 5 days because of gastro-
intestinal upset, and four did not take naproxen regularly
(two had adequate pain relief with dologesic alone, one
found naproxen ineffective, and one complained of leg
cramps and blurred vision after taking naproxen). These
patients were excluded from the analysis. The cumulative
rate of gastric ulcers was significantly lower in patients
pretreated with triple therapy (p=0·04).
Discussion
The interaction between H pylori infection and use of
NSAIDs in the pathogenesis of peptic ulcer is unclear.
Most of the available data is from cross-sectional studies
based on chronic NSAID users, with conflicting
results.12–20 Two studies reported a higher rate of H pylori
in NSAID users with gastroduodenal lesions than in
those with normal mucosa.13,20 Others found significantly
more ulcers in NSAID users who were H pylori positive
than in users not infected with H pylori.12,16,19 But these
findings were not confirmed by other investigators.14,15,17,18
Such conflicting findings arise from different study
designs and outcome measures—for example, they either
assessed the rate of H pylori in NSAID users with or
without mucosal damage13,14 or, conversley, the rate of
mucosal damage in NSAID users with and without H
pylori.15–19 Thus, direct comparison of results is difficult.
These cross-sectional studies do not, therefore, provide
definite evidence for or against a link between H pylori
and NSAIDs in the development of peptic ulcers.
Most of the few published prospective trials did not
indicate that H pylori is a risk factor for NSAID-induced
gastroduodenal damage.21–25 Three studies reported that
H pylori did not affect the severity of gastroduodenal
damage after short-term administration of NSAIDs.21–23
However, the data were mostly derived from studies that
involved small numbers of young healthy volunteers. Any
interaction between H pylori and NSAIDs could have
been obscured by the small number of individuals
included with low event rates in low-risk groups. Two
long-term longitudinal studies of chronic NSAID users
977
THE LANCET
gave conflicting results.25,26 Kim and colleagues25 reported
no significant increase in the rate of gastroduodenal
ulcers among chronic NSAID users with H pylori
infection; patients with erosions detected at baseline were
excluded in the study. By contrast, Taha and colleagues26
found that patients with H pylori-positive duodenal
erosions were more likely than patients who were not
infected with H pylori to develop ulcers during NSAID
treatment. Selective recruitment of chronic users without
complications may introduce bias, since those susceptible
individuals could have dropped out early because of
NSAID intolerance or complications.1,3
The best evidence of a positive interaction between
H pylori and NSAIDs in the pathogenesis of ulcer would
be obtained by investigating the effects of H pylori
eradication on the occurrence of NSAID-induced ulcers.
Seppala and co-workers27 observed that gastric ulcer
relapse seemed to be reduced in a small group of NSAID
users after eradication of H pylori. A prospectve study
by Bianchi and colleagues28 assessed the efficacy of
amoxicillin-omeprazole dual therapy on the healing and
recurrence of ulcers in chronic NSAID users who were
infected with H pylori. They found no significant benefit
although there was a numerical trend towards a higher
rate of ulcer recurrence in patients with H pylori infection.
However, the rate of H pylori eradication was very low,
and their result was further limited by the small number
of patients with ulcer recurrence and the use of different
NSAIDs with variable ulcerogenic potentials.
Our present study is the first prospective randomised
trial to look at whether eradication of H pylori before the
start of NSAID therapy reduces the subsequent risk of
ulcer formation. Our findings establish a pathogenetic
role for H pylori in the development of NSAID-induced
ulcers, in that prophylactic eradication of H pylori
significantly reduced the 8-week cumulative rate of
NSAID ulcers by almost four-fold. The degree of
protection conferred by eradication of H pylori was
similar to that provided by acid-suppression drugs.9,10 The
rate of ulcers in relation to H pylori status was consistent
with our previous findings19 and those reported by
Ekstrom and co-workers.10 Since the risk of ulcer is high
during the first weeks of NSAID therapy, particularly in
those without previous exposure,3,5 our finding is relevant
to patients who need intermittent treatment for
exacerbation of such disorders as osteoarthritis.
Our study differs from others in that most patients in
our series were elderly with concomitant illness—ie,
at high risk of developing NSAID-induced ulcer
complications. We did not recruit patients who were
already on long-term NSAIDs, or those with frequent
previous exposure to NSAIDs, to avoid any potential
selection bias.3,5 All patients received the same dosage of
naproxen to eliminate any confounding effects associated
with different doses and types of NSAIDs.3 In addition,
all endoscopic examinations were done by one
endoscopist who was unaware of treatment allocation to
eliminate bias and between-observer variation. The lower
rate of ulcers in patients on antihelicobacter therapy
might be related to the antiulcer effect of bismuth, but
because bismuth was given for only 1 week before the
start of NSAID therapy, this effect was kept to a
minimum.
Our results show that NSAID-induced ulceration can
be reduced by eradication of H pylori before NSAID
administration and establish H pylori infection as a risk
978
factor for NSAID-induced ulcer disease. Coprescription
of antiulcer prophylactic drugs can reduce the rate of
NSAID ulcers, but the high cost of such treatment
prevents its widespread use. Although it would be costly
and impractical to carry out endoscopy on every patient
before administration of NSAID, H pylori status can be
easily determined by an office-based serology test.
Determination of H pylori and eradication in infected
patients should be recommended before the start of
NSAID therapy.
Contributors
Francis K L Chan, Joseph J Y Sung, S C Sydney Chung, and Jean Woo
were responsible for the original design of the study. Francis K L Chan,
Y T Lee, Vincent K S Leung, and Cynthia S Y Chan coordinated the
study. Data collection, statistical analysis, and independent outcome
assessment were conducted by Francis K L Chan, K F To, M Y Yung,
and Vincent K S Leung. The manuscript was edited by Joseph J Y Sung,
S C Sydney Chung, Edmund K M Li, and Jean Woo. All the authors
contributed to the execution of the study and the writing of the paper.
Acknowledgments
We thank W K Leung, Henry L Y Chan, Justin Wu, W Y So, Teresa
Wong, Gabriel Yip, Thomas S T Li, and Lily Saw for recruitment of
patients and the nursing staff of the Endoscopy Centre, Prince of Wales
Hospital, for their generous support.
References
1 Langman MJS. Anti-inflammatory drugs and the gut-ulcerative
damage and protection from cancer. Excerpta Medica 1995; 72: 1–8.
2 Fries JF, Williams CA, Bloch DA, Michel BA. Nonsteroidal anti-
inflammatory drug-associated gastropathy: incidence and risk factors
model. Am J Med 1991; 91: 213–22.
3 Langman MJS, Weil J, Wainwright P, et al. Risks of bleeding peptic
ulcer associated with individual non-steroidal anti-inflammatory drugs.
Lancet 1994; 343: 1075–78.
4 Hawkey CJ. Non-steroidal anti-inflammatory drug gastropathy: causes
and treatment. Scand J Gastroenterol 1996; 31 (suppl 220): 124–27.
5 Gabriel SE, Jaakkimainen L, Bombardier C. Risks for serious
gastrointestinal complications related to the use of non-steroidal anti-
inflammatory drugs: a meta-analysis. Ann Intern Med 1991; 115:
787–96.
6 Somerville K, Faulkner G, Langman MJS. Non-steroidal anti-
inflammatory drugs and bleeding peptic ulcer. Lancet 1986; i: 452–54.
7 Bloom BS. Direct medical costs of disease and gastrointestinal side
effects during treatment of arthritis. Am J Med 1984; 84 (suppl 2A):
20-24.
8 Silverstein FE, Graham DY, Senior JR, et al. Misoprostol reduces
serious gastrointestinal complications in patients with rheumatoid
arthritis receiving nonsteroidal antiinflammatory drugs: a randomized,
double-blind, placebo-controlled trial. Am Intern Med 1995; 123:
241–49.
9 Taha AS, Hudson N, Hawkey CJ, et al. Famotidine for the prevention
of gastric and duodenal ulcers caused by nonsteroidal
antiinflammatory drugs. N Engl J Med 1996; 334: 1435–39.
10 Ekstrom P, Carling L, Wetterhus S, et al. Prevention of peptic ulcer
and dyspeptic symptoms with omeprazole in patients receiving
continuous non-steroidal anti-inflammatory drug therapy: a Nordic
multicentre study. Scand J Gastroenterol 1996; 31: 753–58.
11 Levine JS. Misoprostol and nonsteroidal anti-inflammatory drugs: a
tale of effects, outcomes, and costs. Ann Intern Med 1995; 123:
309–10.
12 Taha AS, Nakshabendi I, Lee FD, Sturrock RD, Russell RI. Chemical
gastritis and Helicobacter pylori related gastritis in patients receiving
non-steroidal anti-inflammatory drugs: comparison and correlation
with peptic ulceration. J Clin Pathol 1992; 45: 135–39.
13 Heresbach D, Raoul JL, Bretagne JF, et al. Helicobacter pylori: a risk
and severity factor of non-steroidal anti-inflammatory drug induced
gastropathy. Gut 1992; 33: 1608–11.
14 Upadhyay R, Howatson A, McKinlay A, Danesh BJZ, Sturrock RD,
Russell RI. Campylobacter pylori associated gastritis in patients taking
non-steroidal anti-inflammatory drugs. Br J Rheumatol 1988; 27:
113–16.
15 Graham DY, Lidsky MD, Cox AM, et al. Long-term non-steroidal
antiinflammatory drug use and Helicobacter pylori infection.
Gastroenterology 1991; 100: 1653–57.
16 Martin DF, Montgomery E, Dobek AS, Patrissi GA, Peura DA.
Campylobacter pylori NSAIDs, and smoking: risk factors for peptic
ulcer disease. Am J Gastroenterol 1989; 84: 1268–72.
Vol 350 • October 4, 1997

17 Shallcross TM, Rathbone BJ, Wyatt JI, Heatley RV. Helicobacter pylori
associated chronic gastritis and peptic ulceration in patients taking
non-steroidal anti-inflammatory drugs. Aliment Pharmacol Ther 1990;
4: 515–22.
18 Loeb DS, Talley NJ, Ahlquist DA, Carpenter HA, Zinsmeister AR.
Long-term non-steroidal anti-inflammatory drug use and
gastroduodenal injury: the role of Helicobacter pylori. Gastroenetrology
1992; 102: 1899–905.
19 Li EKM, Sung JY, Suen R, et al. Helicobacter pylori infection increases
the risk of peptic ulcers in chronic users in chronic of non-steroidal
anti-inflammatory drugs. Scand J Rheumatol 1996; 25: 42–46.
20 Publig W, Wustinger C, Zandl C. Non-steroidal anti-inflammatory
drugs (NSAIDs) cause gastrointestinal ulcers mainly in Helicobacter
pylori carriers. Wien Klin Wochenschr 1994; 106: 276–79.
21 Lanza FL, Evans DG, Graham DY. Effect of Helicobacter pylori
infection on the severity of gastroduodenal mucosal injury after acute
administration of naproxen or aspirin to normal volunteers. Am J
Gastroenterol 1991; 86: 735–37.
22 Thillainayagam AV, tabaqchali S, warrington SJ, Farthing MJG.
Interrelationships between Helicobacter pylori infection, non-steroidal
anti-inflammatory drugs and gastroduodenal disease: a prospective
study in healthy volunteers. Dig Dis Sci 1994; 39: 1085–89.
Risk of ocular hypertension or open-angle glaucoma in elderly
patients on oral glucocorticoids
Edeltraut Garbe, Jacques LeLorier, Jean-Francois Boivin, Samy Suissa
Summary
Background Ocular hypertension and open-angle glaucoma
are well-known side-effects of treatment with topical
ophthalmic glucocorticoids. There is uncertainty about the
risk of these disorders with oral glucocorticoid therapy.
Methods Data from the Quebec universal health insurance
programme for the elderly were used to identify 9793
patients with a new diagnosis of ocular hypertension or
open-angle glaucoma, or on newly prescribed treatment for
these disorders (cases). 38 325 controls were randomly
selected from ophthalmology patients seen in the same
month and year as the case (index date). Current use of
oral glucocorticoids was defined as that within 14 days of
the index date. All glucocorticoid doses were converted to
the equivalent amount of hydrocortisone. The case-control
analysis was done by conditional logistic regression and
adjusted for age, sex, systemic hypertension, diabetes
mellitus, ophthalmic glucocorticoids, glucocorticoid
injections, and variables related to general health.
Findings The mean ages of cases and controls were similar
(74·9 [SD 6·3] vs 74·7 [6·4]). The adjusted odds ratio of
ocular hypertension or open-angle glaucoma for current
users of oral glucocorticoids compared with non-users was
1·41 (95% CI 1·22–1·63). There was a dose-related
increase in the adjusted odds ratios for current users: 1·26
Division of Clinical Epidemiology, Royal Victoria Hospital
(E Garbe MD, S Suissa PhD); Department of Epidemiology and
Biostatistics, McGill University (E Garbe, J-F Boivin ScD, S Suissa);
Centre de Recherche, Hôtel-Dieu de Montréal, Université de
Montréal, Montreal, Canada (E Garbe, J LeLorier PhD); and
Potsdam Institute of Pharmacoepidemiology and Technology
Assessment, Potsdam, Germany (E Garbe)
Correspondence to: Dr Samy Suissa, Division of Clinical
Epidemiology, Royal Victoria Hospital, 687 Pine Avenue West,
Ross 4.29, Montreal, Quebec H3A 1A1, Canada
(e-mail: SAMYS@EPID.LAN.MCGILL.CA)
Vol 350 • October 4, 1997
THE LANCET
23 Laine L, Cominelli F, Sloane R, Casini-Raggi V, Marin-Sorensen M,
Weinstein WM. Interaction of NSAIDs and Helicobacter pylori on
gastroduodenal injury and prostaglandin production: a controlled
double-blind trial. Ailment Pharmacol Ther 1995; 9: 127–35.
24 Goggin PM, Collins DA, Jazrawi RP, et al. Prevalence of Helicobacter
pylori infection and its effects on symptoms and non-steroidal anti-
inflammatory drug induced gastrointestinal damage in patients with
rheumatoid arthritis. Gut 1993; 34: 1677–80.
25 Kim JG, Graham DY, The Misoprostol Study group. Helicobacter
pylori infection and the development of gastric or duodenal ulcer in
arthritic patients receiving chronic NSAID therapy. Am J Gastroenterol
1994; 89: 203–07.
26 Taha AS, Sturrock RD, Russell RI. Mucosal erosions in longterm non-
steroidal anti-inflammatory drug users: predisposition to ulceration
and relatiom to Helicobacter pylori. Gut 1995; 36: 334–36.
27 Seppala K, Pikkarainen P, Sipponen P, Kivilaakso E, Gormsen MH,
and the Finnish Gastric Ulcer Study Group. Cure of peptic ulcer
associated with eradication of Helicobacter pylori. Gut 1995; 36: 834–37.
28 Bianchi Porro G, Parente F, Imbesi V, Montrone F, Caruso I. Role of
Helicobacter pylori in ulcer healing and recurrence of gastric and
duodenal ulcers in longterm NSAID users: response to omeprazole
dual therapy. Gut 1996; 59: 22–26.
(1·01–1·56) for less than 40 mg per day of hydrocortisone,
1·37 (1·06–1·76) for patients on 40–79 mg per day, and
1·88 (1·40–2·53) for patients on 80 mg or more per day.
The odds ratios also increased with the duration of
treatment over the first 11 months of exposure.
Interpretation The use of oral glucocorticoids increases the
risk of ocular hypertension or open-angle glaucoma in
elderly patients. In patients in this age-group who need
long-term treatment with high doses of oral
glucocorticoids, monitoring of intraocular pressure may be
justified.
Lancet 1997; 350: 979–82
Introduction
Ocular hypertension is a well-known side-effect of topical
ophthalmic glucocorticoids.1 The lack of symptoms with
raised intraocular pressure has prompted ophthalmol-
ogists to recommend routine monitoring of intraocular
pressure in patients on long-term ophthalmic steroids.1,2
Several case reports have suggested that oral
glucocorticoid therapy can also result in ocular
hypertension and open-angle glaucoma.3-5 However,
investigations into the effects of oral corticosteroids on
ocular pressure are scarce. There have been a few
retrospective studies with small numbers of patients on
glucocorticoids for variable periods of time.6–10 In some of
these studies the mean intraocular pressure was higher in
the steroid-treated group than in the control group.6,7
Most ophthalmologists believe that the risk of ocular
hypertension or open-angle glaucoma is lower with
oral glucocorticoids than with topical ophthalmic
glucocorticoids, although the risk has not yet been
quantified.11,12 Some investigators have hypothesised that
months or years of treatment with oral glucocorticoids are
needed to cause ocular hypertension. By contrast, most
clinicians recognise an increased risk with topical
ophthalmic steroids after weeks of treatment.11,13
979