Full text of DOI:10.2165/00019053-200119080-00005

Pharmacoeconomics 2001; 19 (8): 831-843
1170-7690/01/0008-0831/$22.00/0
ORIGINAL RESEARCH ARTICLE
©
Adis International Limited. All rights reserved.
Cost Effectiveness in Canada of a
Multidrug Prepackaged Regimen
®
†
(
Hp-PAC ) for Helicobacter pylori
Eradication
Karen Agro, Gordon Blackhouse, Ron Goeree,1,2 Andrew R. Willan,1,2 Jia-Qing
1
1,2
3
3
1,2
Huang, Richard H. Hunt and Bernie J. O’Brien
1
2
3
Centre for Evaluation of Medicines, St Joseph’s Hospital, Hamilton, Ontario, Canada
Department of Clinical Epidemiology, McMaster University, Hamilton, Ontario, Canada
Department of Gastroenterology, McMaster University, Hamilton, Ontario, Canada
Abstract
Objective: To assess the cost effectiveness of a multidrug prepackaged regimen
®
for Helicobacter pylori, the Hp-PAC (lansoprazole 30mg, clarithromycin 500mg,
amoxicillin 1g, all twice daily), relative to alternative pharmacological strategies
in the management of confirmed duodenal ulcer over a 1-year period from 2
perspectives: (i) a strict healthcare payer perspective (Ontario Ministry of Health)
excluding the patient copayment; and (ii) a healthcare payer perspective including
the patient copayment.
Design: A decision-analytical model was developed to estimate expected per
patient costs [1998 Canadian dollars ($Can)], weeks without ulcer and symptom-
®
atic ulcer recurrences for the Hp-PAC compared with: proton pump inhibitor
(
(
PPI)-clarithromycin-amoxicillin (PPI-CA), PPI-clarithromycin-metronidazole
PPI-CM), PPI-amoxicillin-metronidazole (PPI-AM) and ranitidine-bismuth-
metronidazole-tetracycline (RAN-BMT).
Main outcome measures and results: All PPI-based regimens had higher ex-
pected costs but better outcomes relative to RAN-BMT. From a strict healthcare
payer perspective, PPI-CM ($Can209) yielded lower expected costs than PPI-CA
®
(
$Can221) and slightly lower costs than Hp-PAC ($Can211). However, these 3
regimens all shared identical outcomes (51.2 weeks without ulcer). When the
current Ontario, Canada, $Can2 patient copayment was added to the dispensing
®
fee, Hp-PAC yielded lower costs ($Can214) than PPI-CM ($Can216).
®
Conclusion: From a strict healthcare payer perspective, Hp-PAC is weakly
dominated by PPI-CM with an incremental cost effectiveness (relative to RAN-
BMT) of $Can5.77 per ulcer week averted. When the patient copayment is added
®
to this perspective, Hp-PAC weakly dominates PPI-CM ($Can5 per ulcer week
®
averted). Regardless of perspective, Hp-PAC and PPI-CM differed by only
$
Can2 per patient over 1 year and the expected time without ulcer was 51.2 weeks
for both. More data on the clinical and statistical differences in H. pylori eradi-
®
cation with Hp-PAC and PPI-CM would be useful. This analysis does not in-
†
Use of tradenames is for product identification only and does not imply endorsement.

8
32
Agro et al.
®
clude the possible advantage of Hp-PAC in terms of compliance and antibacte-
rial resistance.
Helicobacter pylori is recognised as a major
In a recent economic evaluation, Taylor et al.[6]
found bismuth triple therapy and triple therapies
containing PPIs to be associated with the lowest
rate of ulcer recurrence and cost. Although Taylor
and colleagues focused on recent clinical data and
new eradication regimens, there were limitations
with the analysis. First, efficacy rates were ad-
justed using 1987 compliance data from a nonster-
oidal anti-inflammatory drug study. Secondly, ul-
cer rates were not adjusted for ‘symptomatic’
recurrence. Hence, the recurrence rates may be an
overestimate of what would be observed in clinical
practice. Finally, data from all PPIs were combined
without first assessing if they produce the same
rates of healing.
contributor to the development of duodenal ulcer.
A1994 US National Institutes of Health (NIH) con-
sensus conference noted that ‘. . . nearly all pa-
tients with duodenal ulcer have H. pylori’, and that
the association with gastric ulcer is ‘. . . only
[
1]
slightly less strong’. The NIH panel recom-
mended treatment with antimicrobial agents for pa-
tients with peptic ulcer disease with H. pylori in
addition to antisecretory drugs whether on first pre-
sentation with the illness or on recurrence.
Recent Canadian consensus guidelines empha-
sise the need for H. pylori eradication and suggest
proton pump inhibitor (PPI)-clarithromycin-
amoxicillin (PPI-CA) and PPI-clarithromycin-
metronidazole (PPI-CM) as first-line eradication
Recently, the Canadian Therapeutic Products
Directorate approved a pre-packaged PPI-contain-
regimens because of their high eradication rates
2]
®
(
≥80%).[
ing triple regimen called the Hp-PAC (lansoprazole
A number of economic evaluations have been
30mg, clarithromycin 500mg, amoxicillin 1g, all
twice daily). An economic advantage of the Hp-
conducted since the NIH consensus conference.
O’Brien et al.[ structured the economic problem
as a comparison between 3 global treatment strat-
egies: (i) intermittent acid suppression [H2 receptor
antagonist (H2RA) or PPI]; (ii) maintenance acid
suppression (H2RA); or (iii) H. pylori eradication
with an H2RA and bismuth triple therapy or with
omeprazole plus amoxicillin or clarithromycin.
Eradication strategies were less costly and more
effective (fewer symptomatic recurrences) than the
other 2 strategies; however, because of limited
studies at the time, they could not conclude which
of the H. pylori regimens offered best value for
money.
3]
®
PAC is that it will be dispensed as a single prescrip-
tion attracting one dispensing fee. Traditionally, the
individual medications would be dispensed as sepa-
rate prescriptions each with a dispensing fee.
Therefore, the objective of this study is to com-
pare the expected costs and outcomes of the Hp-
®
PAC relative to currently used PPI-based regimens
and ranitidine-bismuth-metronidazole-tetracycline
(RAN-BMT) from 2 perspectives: (i) a strict health-
care payer perspective [Ontario Ministry of Health
(MOH)] excluding the patient copayment, hereaf-
ter referred to as MOH+; and (ii) a healthcare payer
perspective including the patient copayment, here-
after referred to as MOH–. Considering both per-
spectives is especially important for this problem
because a traditional 3-drug regimen for H. pylori
would require a patient in Ontario, Canada, to pay
3 copayments of 2 Canadian dollars ($Can) for
each drug.
In a subsequent evaluation for the Canadian Co-
ordinating Office of Health Technology Assess-
[
4,5]
ment (CCOHTA), O’Brien et al.
confirmed that
eradication strategies were dominant over intermit-
tent or maintenance acid suppression. Among the
H. pylori eradication strategies, triple therapy with
omeprazole and metronidazole and either clar-
ithromycin or amoxicillin offered the best value for
money over bismuth triple therapy.
Given that the active ingredients of Hp-PAC®
are the same as PPI-CA, a secondary objective of
the study was to determine the economic impact of
©
Adis International Limited. All rights reserved.
Pharmacoeconomics 2001; 19 (8)

Prepackaged Regimen for H. pylori Eradication
833
multiple drug pre-packaging on drug and dispens-
ing costs. Therefore, we attempt to answer the ques-
tion: how will the perspective of a cost-effectiveness
analysis affect the rank ordering of the Hp-PAC®
relative to other H. pylori eradication regimens?
based eradication regimen, we assumed that they
receive 2 weeks of a PPI (administered on a twice
daily regimen). If they remain unhealed at 4 weeks,
they receive another 2 weeks of PPI monotherapy.
At 6 weeks, we assumed all patients are healed. If
patients are unhealed following RAN-BMT (where
ranitidine is given for 4 weeks), we assumed they
receive another 4 weeks of ranitidine at which
point we assumed 100% healing. The probability
of ulcer relapse in subsequent 3 month cycles is
dependent upon whether H. pylori was success-
fully eradicated by the treatment regimen.
Methods
Overview of Analytical Approach
The analytical approach was divided into 3 key
components. First, the therapeutic decision prob-
lem was structured using the principles of clinical
decision analysis where clinical events and costs
are structured as a decision tree. Secondly, ex-
pected costs and outcomes were calculated by de-
termining the probabilities of relevant clinical
events (i.e. ulcer healing rates, H. pylori eradica-
tion rates, ulcer recurrence rates) using principles
of data synthesis from reviewed literature. Search
criteria were identified and used for retrieval and
subsequent pooling of relevant rates from publish-
ed studies for inclusion in the model. Thirdly, prin-
ciples of cost-effectiveness analysis were used for
comparing treatment strategies in terms of weak
and strong dominance and incremental cost effec-
tiveness. Sensitivity analysis was used to explore
key areas of uncertainty.
If patients relapse, we assumed they will be
treated with a different drug regimen. Patients first
treated with PPI-amoxicillin-metronidazole (PPI-
AM), PPI-CM and RAN-BMT strategies are as-
sumed to be switched to Hp-PAC® upon recur-
rence. Patients first treated with PPI-CA and
®
Hp-PAC are assumed to be switched to PPI-CM
treatment upon recurrence. For second and third
relapses, we assumed patients will be treated with
a PPI to heal followed by maintenance PPI therapy.
Clinical and Cost Outcome Measures
Our primary clinical outcome is ‘time free from
ulcer’over the 12-month period of the model. This
outcome measure incorporates both the speed with
which an ulcer is healed and the likelihood of an
ulcer recurrence. The expected costs for each strat-
egy include both initial drug therapy and ‘down-
stream’ costs due to management of ulcer recur-
rence in the 1-year interval. The model does not
assume additional diagnostic testing for the pres-
ence of H. pylori or testing to confirm eradication.
Decision-Analysis Model
Our model was based on a Markov process with
a cycle length of approximately 3 months (exactly
9
1 days). A Markov model is a recursive process
that is useful for clinical situations where patients
have an ongoing risk of experiencing a defined set
of events.[ The timing of these events may be un-
certain and the event may occur more than once. A
7]
3
-month Markov cycle model allows for the possi-
Key Clinical Model Inputs
bility of more than 2 ulcer relapses per year. Three
months represents a reasonable period of time over
which relapses can be observed.
The model commences with a cohort of patients
with confirmed, uncomplicated duodenal ulcer ini-
tially receiving ‘heal-and-eradicate’therapy with 1
of the 5 treatment strategies under study as outlined
in table I. If patients are unhealed following a PPI-
There were 3 key clinical inputs into the deci-
sion model: time with ulcer per healing episode; H.
pylori eradication rates for each strategy; and the
probability of ulcer recurrence. These parameters
are discussed in the following sections. All costs
are expressed in 1998 Canadian dollars ($Can)
[$US1 = $Can1.48].
©
Adis International Limited. All rights reserved.
Pharmacoeconomics 2001; 19 (8)

8
34
Agro et al.
Table I. Costs of drug regimens, excluding dispensing fee
Regimen
Dose
Days of therapy
from-to)
Cost ($Can per drug)a,b
Cost ($Can per regimen)
(
PPI-AM
Average PPI
c
1-7
1-7
1-7
1-7
1-7
1-7
31.36
6.16
Amoxicillin
[500mg × 2] bid
[250mg × 2] bid
20mg bid
Metronidazole
0.84
38.36
40.68
37.80
36.26
PPI-AM (omeprazole as PPI)
PPI-AM (lansoprazole as PPI)
PPI-AM (pantoprazole as PPI)
33.88
30.80
29.26
30mg bid
40mg bid
PPI-CA
Average PPI
c
1-7
1-7
1-7
1-7
1-7
1-7
31.36
45.50
6.16
Clarithromycin
500mg bid
[500mg × 2] bid
20mg bid
Amoxicillin
83.02
85.54
82.46
80.09
PPI-CA (omeprazole as PPI)
PPI-CA (lansoprazole as PPI)
PPI-CA (pantoprazole as PPI)
33.88
30.80
29.26
30mg bid
40mg bid
PPI-CM
Average PPI
c
1-7
1-7
1-7
1-7
1-7
1-7
31.36
45.50
0.84
Clarithromycin
500mg bid
[250mg × 2] bid
20mg bid
Metronidazole
77.70
80.22
77.14
75.74
PPI-CM (omeprazole as PPI)
PPI-CM (lansoprazole as PPI)
PPI-CM (pantoprazole as PPI)
33.88
30.80
29.26
30mg bid
40mg bid
RAN-BMT
Ranitidine
Bismuthd
150mg bid
1-28
1-14
1-14
1-14
24.64
16.80
1.68
2 tablets qid
250mg qid
Metronidazole
Tetracycline
[250mg × 2] qid
2.24
45.36
Hp-PAC®
Lansoprazole
30mg bid
1-7
1-7
1-7
Clarithromycin
500mg bid
Amoxicillin
[500mg × 2] bid
80.96
Maintenance PPI
c
Average PPI
1-91
203.84
203.84
a
b
Unit price sources: all Ontario Drug Benefit prices (1998), except bismuth (survey of local pharmacies) and clarithromycin 500mg (manu-
facturers price with 10% pharmacy mark-up), all with mark-up.
Unit prices (all with 10% mark-up excluding dispensing fees): ranitidine (generic) 150mg, $Can0.44; omeprazole 20mg, $Can2.42;
®
lansoprazole 30mg, $Can2.20; pantoprazole 40mg, $Can2.09; Pepto-Bismol tablets, $Can0.15; amoxicillin (generic) 500mg,
$Can0.22; clarithromycin 500mg, $Can3.25; metronidazole (generic) 250mg, $Can0.03; tetracycline 250mg, $Can0.02.
c
Omeprazole 20mg, lansoprazole 30mg, pantoprazole 40mg.
®
d
Pepto-Bismol contains 262mg bismuth subsalicylate and 350mg calcium carbonate; 2 tablets of bismuth suggested in recent publica-
tions.[
8]
bid = twice daily; Hp-PAC® = lansoprazole-clarithromycin-amoxicillin, prepackaged; PPI = proton pump inhibitor; PPI-AM = PPI-amoxicillin-
metronidazole; PPI-CA = PPI-clarithromycin-amoxicillin; PPI-CM = PPI-clarithromycin-metronidazole; qid = 4 times daily; RAN-BMT =
ranitidine-bismuth-metronidazole-tetracycline; $Can = Canadian dollars.
©
Adis International Limited. All rights reserved.
Pharmacoeconomics 2001; 19 (8)

Prepackaged Regimen for H. pylori Eradication
835
Time with Ulcer per Healing Episode
mark-up. For drugs not listed on the formulary, ei-
ther the manufacturer’s price with a 10% mark-up
was used (for clarithromycin 500mg) or a small
survey of local pharmacies was conducted (for
Using random effects meta-analysis tech-
niques,[
9-11]
the probability of healing was esti-
mated at different time points using pooled data
from published studies. In all PPI-based ulcer heal-
ing studies, all patients had endoscopically diag-
nosed duodenal ulcer at baseline. H. pylori positiv-
ity was assessed by at least 1 method in the PPI
plus 1 antibacterial and PPI plus 2 antibacterial
studies. Studies for PPI-based treatments were pri-
marily identified from Medline (1985 to May
®
Pepto-Bismol ). Regimen costs (with the 10%
mark-up) are presented in table I along with varia-
tions for sensitivity analyses in the regimen costs
when different PPIs are used in PPI-AM, PPI-CA
and PPI-CM. The PPI cost used in PPI-AM, PPI-
CAand PPI-CM, and maintenance PPI therapy was
a simple average of the 3 PPIs (omeprazole,
lansoprazole, pantoprazole).
In Ontario, pharmacies dispensing an ODB eli-
gible prescription are entitled to a $Can6.11 dis-
pensing fee. Aportion of this fee, $Can4.11, is paid
by the ODB programme directly to the pharmacy.
The patient pays the balance of the fee, the
[
4]
1
997) and a previous economic analysis. Ranit-
idine healing data were estimated from an earlier
analysis.[
4]
Helicobacter pylori Eradication Probabilities
For PPI-based regimens, no published meta-
analysis on eradication rates was available that
summarised all regimens of interest in the duode-
nal ulcer patient population. Therefore, we esti-
mated the eradication rates based on data provided
$
Can2.00 copayment. For the MOH+ perspective,
a dispensing fee of $Can4.11 was used ($Can6.11
ODB fee minus $Can2.00 patient copayment). For
the MOH– perspective, patients pay a $Can2.00
copayment (total $Can6.11). Other out-of-pocket
costs, such as over-the-counter antacids, were not
included but there is no reason to suppose these
costs would vary between regimens. Only dispens-
ing fees and copayments set by the MOH are con-
sidered in this analysis. Across Canada, dispensing
fees and copayments vary widely depending on the
type of payer (i.e. public, private or cash-paying
[
12]
in a recent preliminary report
and a threshold
value provided from the Canadian H. pylori con-
sensus conference.[ For RAN-BMT, we utilised
2]
[
13]
an earlier randomised controlled trial and meta-
analysis.[
14]
Ulcer Recurrence Probabilities
Studies reporting ulcer recurrence rates based
on H. pylori status were identified and analysed
using literature sources similar to those for the
healing analysis [Medline (1992 to 1997) and a pre-
patient).
The Hp-PAC® cost was determined from the
[
4]
vious economic analysis ]. Recurrence rates were
adjusted to account for the fact that approximately
manufacturer to be $Can73.60 ($Can80.96 incor-
porating the 10% pharmacy mark-up). From the
7
6% of endoscopically diagnosed ulcer recur-
®
MOH+ perspective, the Hp-PAC attracts one dis-
[
4]
rences are symptomatic, resulting in a physician
visit.
pensing fee ($Can4.11), and therefore its total cost
is $Can85.07 ($Can9.72 less than the individually
dispensed drugs each with a dispensing fee). From
the MOH– perspective, a $Can6.11 fee is charged
Key Cost Model Inputs
(
$Can4.11 paid by the MOH and $Can2 by the pa-
In the decision model, costs are incurred for: (i)
drug therapy for the initial duodenal episode; and
ii) the management of recurrent duodenal ulcers.
tient), and therefore its total cost is $Can87.07.
(
Cost of Managing Duodenal Ulcer Recurrences
Resource use data associated with the manage-
ment of symptomatic ulcer recurrences were de-
rived from a previous expert physician panel (4
gastroenterologists, 2 family doctors),[ where a
Cost of Treating Duodenal Ulcer Episodes
The primary source of drug price information
was the Best Available Price from the Ontario Drug
Benefit (ODB) Formulary with a 10% pharmacy
4]
©
Adis International Limited. All rights reserved.
Pharmacoeconomics 2001; 19 (8)

8
36
Agro et al.
modified Delphi technique[15] was used to estimate
the percentage likelihood and volume of various
services used when patients present with ulcer re-
currence symptoms.
Sensitivity Analyses
To assess the impact of uncertainty surrounding
many of the data estimates, sensitivity analyses
were conducted on the: (i) PPI-AM eradication
rate; (ii) eradication rate of PPI-CM versus PPI-CA
to simulate metronidazole and/or clarithromycin
resistance; (iii) cost of omeprazole, lansoprazole
and pantoprazole in the PPI-based regimens; and
(iv) effect of increasing the dispensing fee.
In this analysis, we have used the physician pan-
el’s recommendations with the exception of gastro-
enterologist visits in an attempt to reflect changes
in the management of H. pylori. Originally, 1 gas-
troenterologist reassessment and 2 gastroenterolo-
gist partial assessments were assumed for each re-
currence. For this analysis, we assumed the general
practitioner (GP) will conduct the follow-up and
thus have removed the gastroenterologist partial
assessments. A GP minor assessment is now as-
sumed if patients are unhealed following H. pylori
eradication therapy.
Results
Clinical Model Input Results
Ulcer Healing Probabilities and Time-with-Ulcer
per Healing Episode
The proportion of patients healed at various
time points for PPI alone, PPI plus 1 antibacterial,
PPI plus 2 antibacterials and ranitidine-based re-
gimens are provided in table III and figure 1. From
these data, 3 separate healing curves were esti-
mated with simple linear interpolation between
time points. The time-with-ulcer per healing epi-
sode was calculated as the area above each curve.
The time-with-ulcer per healing episode was ap-
plied to duodenal ulcer episodes treated with PPI
plus 2 antibacterial regimens, whereas the time-
with-ulcer for PPI alone was applied to duodenal
Costs (combined hospital and physician service
cost) for procedures such as upper gastrointestinal
endoscopy were estimated from 2 sources: (i) an
Ontario Case Costing Project hospital in South-
Western Ontario; and (ii) the physician fee sched-
ule for Ontario,[ which itemises allowable phy-
sician reimbursement by procedure under the
provincial health insurance plan. These unit costs,
physician fees and utilisation estimates by recur-
rence are provided in table II.
16]
Table II. Healthcare utilisation and costs of managing ulcer recurrence (excluding drugs)a
Cost component
Unit cost ($Can)
First recurrence
%
Second and subsequent recurrences
cost ($Can)
38.65
%
cost ($Can)
Visits to gastroenterologist
Gastroenterologist reassessment
38.65
1b
1b
38.65
Tests and procedures
Abdominal ultrasound (complete)
Gastric biopsy
Complete blood count
Culture (Helicobacter pylori)
Serum gastrin
99.61
325.00
9.21
46.69
7.72
0
2
2.00
308.75
4.61
23.35
0.39
95
50
0
308.75
4.61
95
50
50
5
0
Upper gastrointestinal endoscopy
Urea ($Can5.98) + electrolytes
118.22
20.22
95
25
112.31
5.06
95
40
112.31
8.09
($Can14.24)
Cost per recurrence
469.38
498.15
a
b
$
Modified from the Canadian Coordinating Office of Health Technology Assessment (CCOHTA) report.[4]
Number of visits.
Can = Canadian dollars.
©
Adis International Limited. All rights reserved.
Pharmacoeconomics 2001; 19 (8)

Prepackaged Regimen for H. pylori Eradication
837
Table III. Cumulative probability of ulcer healing by weeks on therapy
Regimen
1 Week
2 Weeks
4 Weeks
8 Weeks
number of probability
study arms healed (95%
CI)
number of probability
study arms healed (95%
CI)
number of probability
study arms healed (95%
number of probability
study arms healed (95%
CI)
CI)
0.89
a
PPI alone
43
16
7
0.64
0.61, 0.68)
0.81
(0.75, 0.86)
0.94
55
12
14
26
(
(0.87, 0.90)
0.92
PPI plus 1
antibacterial
b,c
(0.90, 0.95)
e
PPI plus 2
antibacterials
11
0.92
(0.88, 0.95)
0.96
c,d
(0.90, 0.98)
(0.94, 0.98)
0.74
Ranitidine
25
0.88f
(0.69, 0.78)
(0.85, 0.91)
a
b
c
d
e
f
Area-above-the-curve for 4 weeks of PPI alone = 1.83 weeks with ulcer.
PPI plus amoxicillin and PPI plus clarithromycin studies found. Studies analysed together.
Due to large variations in antibacterial dose and duration, all doses analysed together.
All antibacterial types analysed together.
Area-above-the-curve for 4 weeks of a PPI plus 2 antibacterials = 0.71 weeks with ulcer.
Area-above-the-curve for 8 weeks of ranitidine = 3.30 weeks with ulcer.
CI = confidence interval; PPI = proton pump inhibitor.
ulcer episodes treated with a maintenance PPI reg-
imen.
Area-above-the-curve calculations for 4 weeks
of a PPI alone resulted in 1.83 weeks with ulcer.
Similarly, 4 weeks of a PPI plus 2 antibacterials
and 8 weeks of ranitidine resulted in 0.71 and 3.30
weeks with ulcer, respectively.
interval (CI) 1.0, 5.0; 19 study arms] and at 12
months was 2.0% (95% CI 1.0, 3.0; 28 arms). The
12-month ulcer recurrence rate may have been
lower than the 6-month rate because of more stud-
ies reporting 12-month data. Therefore, in the
model we assumed 2.5% as the 12-month duodenal
ulcer recurrence rate (the average between the 6-
and 12-month values). In H. pylori–positive pa-
tients the 6-month recurrence rate was 52% (95%
CI 35, 69; 20 arms) and the 12-month recurrence
rate was 62% (95% CI 45, 80; 21 arms).
For maintenance PPI therapy, we assumed that
the ulcer recurrence rate would approximate that
of the H. pylori–negative rate because the patient
is likely to be H. pylori–negative following 2 H.
pylori eradication regimens and maintenance PPI
therapy. All recurrence rates were adjusted by a
factor of 0.76 to approximate symptomatic ulcer
recurrences.
H. pylori Eradication Probabilities
For PPI-based regimens, the following eradica-
tion rates were used in the model: PPI-AM 71%,
PPI-CA 88% and PPI-CM 90%. The PPI-CA and
PPI-CM rates were consistent with those reported
by Huang et al.,[ who examined H. pylori eradi-
cation rates in patients with duodenal ulcer, gastric
ulcer and non-ulcer dyspepsia. These rates are
broadly consistent with the threshold rates re-
ported in the Canadian H. pylori consensus confer-
ence (i.e. PPI-CAand PPI-CM are reported to have
greater than 80% eradication rates on an intention-
to-treat basis).[ For RAN-BMT, 86% was used as
12]
2]
[
14]
Cost-Effectiveness Results
the base case from the meta-analysis by Penston
(
BMT for 2 weeks of therapy).
In tableIV, wepresent both the MOH+ andMOH–
perspectives of our analysis. Our base case is the
MOH+ perspective. For each of the 5 strategies we
present expected 1-year costs per patient, symp-
Ulcer Recurrence Probabilities
In H. pylori–negative patients, the ulcer recur-
rence rate at 6 months was 3.0% [95% confidence
©
Adis International Limited. All rights reserved.
Pharmacoeconomics 2001; 19 (8)

8
38
Agro et al.
1
0
0
0
0
.0
.8
.6
.4
.2
0
PPI plus 2 antibacterials
PPI plus 1 antibacterial
PPI plus no antibacterials
Ranitidine
0
1
2
3
4
5
6
7
8
Time (wks)
Fig. 1. Healing time curves for ranitidine and proton pump inhibitor (PPI)-based regimens (95% confidence interval).
tomatic ulcer recurrences per 100 patients and ex-
pected weeks per patient without ulcer in 1 year.
The rate of symptomatic ulcer recurrence per
BMT. Among the PPI regimens, PPI-AM is strongly
dominated (having higher costs and lower benefits
than an alternative). PPI-CM yields lower expected
costs than PPI-CA ($Can12) and slightly lower
1
00 patients is lowest for the PPI-CM regimen (7.3
®
®
per 100) followed by PPI-CA (8.2) and Hp-PAC
costs than Hp-PAC ($Can2) over a 1-year period.
(
8.2), RAN-BMT (9.4) and PPI-AM (17.2). The
However, these 3 regimens have identical outcomes,
and therefore Hp-PAC® and PPI-CA are only
weakly dominated (i.e. higher cost with same out-
come as alternative).
rank-ordering of regimens according to recurrence
is consistent with prior expectation given the erad-
ication rates used in the model with PPI-CM hav-
ing the highest eradication rate at 90% and PPI-CA
Following accepted conventions for comparing
®
[17]
and the Hp-PAC sharing the same eradication
mutually exclusive strategies,
these dominated
rate. The expected weeks per patient without ulcer
were equivalent at 51.2 weeks for PPI-CM, PPI-CA
alternatives are not considered further in the incre-
mental analysis and an efficiency frontier is the
locus of nondominated alternatives in the incre-
mental cost, incremental effectiveness (ΔC, ΔE)
space relative to RAN-BMT as the origin. From the
MOH+ perspective, PPI-CM is the only non-
dominated alternative on this locus. PPI-CM yields
an incremental cost-effectiveness ratio of $Can5.77
per ulcer week averted relative to RAN-BMT.
From the MOH– perspective, the $Can2 copay-
ment is added to the dispensing fee as an out-of-
pocket expense. Outcomes remain the same but the
expected cost per patient over 1 year changes:
PPI-AM ($Can240), PPI-CA ($Can228), PPI-CM
®
and Hp-PAC and were lower for PPI-AM (51.1
weeks) and for RAN-BMT (48.6 weeks).
From an MOH+ perspective, the expected cost
per patient over 1 year was highest for PPI-AM
(
$Can234) followed by PPI-CA ($Can221), Hp-
®
PAC ($Can211), PPI-CM ($Can209) and RAN-
BMT ($Can194).
The relationships between costs and outcomes
are presented in table IV and the cost and effect
coordinates are plotted in figure 2, with RAN-BMT
used as the reference point. From an MOH+ per-
spective, the PPI-based regimens have higher ex-
pected costs with better outcomes relative to RAN-
®
($Can216), Hp-PAC ($Can214) and RAN-BMT
©
Adis International Limited. All rights reserved.
Pharmacoeconomics 2001; 19 (8)

Prepackaged Regimen for H. pylori Eradication
839
(
$Can201). The rank ordering (by cost) of Hp-
dominated alternative on the efficiency frontier
with an incremental cost effectiveness of $Can5
per ulcer-week averted (fig. 2b).
®
PAC and PPI-CM change as the expected cost of
®
Hp-PAC becomes $Can2 lower than for PPI-CM.
Hp-PAC yields an incremental cost-effectiveness
ratio of $Can5.00 per ulcer week averted relative
to RAN-BMT.
®
Sensitivity Analyses
We conducted 4 main sensitivity analyses. We
varied the: (i) PPI-AM eradication rate; (ii) eradi-
cation rate of PPI-CM versus PPI-CA to simulate
metronidazole and/or clarithromycin resistance;
In summary, RAN-BMT remains the least
costly and least effective strategy from both per-
spectives. However, the choice of perspective de-
termines which strategy is the next nondominated
but more costly and more effective strategy that
should be considered if a decision-maker is willing
to pay more for greater outcomes (i.e. the relevant
mutually exclusive alternative for calculation of an
incremental cost-effectiveness ratio). From an
MOH+ perspective, PPI-CM is the only strategy
on the efficiency frontier with an incremental cost
effectiveness of $Can5.77 per ulcer-week averted
(
dominated by PPI-CM (having same effects and
higher costs), and PPI-AM is strongly dominated
(
(
iii) omeprazole, lansoprazole and pantoprazole
costs in the PPI-based regimens; and (iv) the dis-
pensing fee.
We found that PPI-AM eradication rates range
considerably in the literature. The draft duodenal
[18]
ulcer guidelines from Holbrook et al.
(being
produced for the MOH) suggest a 77% eradication
rate for PPI-AM. Changing the PPI-AM eradica-
tion rate to 77% lowers the expected cost for this
strategy. However, PPI-AM is still dominated by
®
fig. 2a). Both PPI-CA and Hp-PAC are weakly
®
PPI-CM and Hp-PAC .
The eradication rate of PPI-CM was varied to
assess the impact of metronidazole and/or clar-
ithromycin resistance on the cost effectiveness of
having lower effects and higher costs).
If the $Can2 copayment is included (MOH–
perspective), Hp-PAC® becomes the only non-
PPI-CM relative to Hp-PAC . When the eradica-
®
Table IV. Expected costs, ulcer recurrences and incremental cost effectiveness using weeks without ulcer over 1 year
Strategy
Expected cost
per patient
Symptomatic ulcer Expected weeksa
ΔC ($Can)
ΔE (ulcer weeks ΔC/ΔE ($Can/ulcer
recurrences per
100 patients
without ulcer
averted)
week averted)
($Can)
MOH+ perspective (base case)
RAN-BMT
PPI-CM
Hp-PAC®
PPI-CA
194
209
211
221
234
9.4
7.3
8.2
8.2
17.2
48.6
51.2
51.2
51.2
51.1
–
15
–
2.6
–
5.77
Weakly dominated
Weakly dominated
Strongly dominated
PPI-AM
MOH– perspective
RAN-BMT
Hp-PAC®
PPI-CM
PPI-CA
201
9.4
8.2
7.3
8.2
17.2
48.6
51.2
51.2
51.2
51.1
–
13
–
2.6
–
5.00
214
216
228
240
Weakly dominated
Weakly dominated
Strongly dominated
PPI-AM
a
Rounded to the first decimal place.
CI = confidence interval; Hp-PAC® = lansoprazole-clarithromycin-amoxicillin, prepackaged; MOH+ = Ontario Ministry of Health perspective
excluding the patient copayment; MOH– = MOH perspective including the patient copayment; PPI = proton pump inhibitor; PPI-AM =
PPI-amoxicillin-metronidazole; PPI-CA = PPI-clarithromycin-amoxicillin; PPI-CM = PPI-clarithromycin-metronidazole; RAN-BMT = ranitid-
ine-bismuth-metronidazole-tetracycline; $Can = Canadian dollars; ΔC = incremental cost; ΔC/ΔE = incremental cost-effectiveness ratio;
ΔE = incremental effectiveness.
©
Adis International Limited. All rights reserved.
Pharmacoeconomics 2001; 19 (8)

8
40
Agro et al.
®
PPI-AM
except Hp-PAC ). For example, if omeprazole is
substituted as the PPI, it is used in the PPI-AM,
PPI-CA and the PPI-CM regimens. It is also used
as the PPI in the switch therapies and in the PPI
maintenance therapy for all regimens. This is why
PPI-CA
Hp-PAC®
PPI-CM
RAN-BMT
a
MOH+ perspective
5
4
3
2
1
0
0
0
0
®
the expected costs of the Hp-PAC arm change (i.e.
(2.5,40)
lansoprazole is always the PPI in this regimen, but
the PPI in the switch therapy and maintenance PPI
therapy will change within this arm).
(2.6,27)
(
(
2.6,17)
2.6,15)
In our base case, the PPI cost was a simple aver-
age of the 3 PPIs available on the Canadian market.
The average cost of the 3 PPIs was compared with
0
0
(0,0)
®
the Hp-PAC in which the cost of lansoprazole is
b
used (pre-packaged with lansoprazole). When the
individual PPI cost for omeprazole is used, the rank
MOH− perspective
5
4
3
2
1
0
0
0
0
®
order of PPI-CM and Hp-PAC are identical. With
(2.5,39)
®
omeprazole, the expected costs for Hp-PAC and
(
2.6,27)
2.6,15)
PPI-CM are both $Can212. With pantoprazole, the
®
expected costs of Hp-PAC are $Can210 and for
(
PPI-CM are $Can207. The cost rank ordering of
0
0
(2.6,13)
®
PPI-CM and Hp-PAC does vary with respect to
(
0,0)
the assumed PPI acquisition cost, but this variation
is small (±$Can2) with respect to the overall 1-year
cost per patient.
0
1
2
3
Incremental benefits (ulcer weeks averted)
Fig. 2. Incrementalcost-effectiveness(CE)relationships between
the 5 treatment regimens. (a) MOH+ perspective: CE ratio (PPI-
CM vs RAN-BMT) = 5.77; (b) MOH– perspective: CE ratio
In figure 3, we display the effect of increasing
the dispensing fee per item on the expected costs
®
Hp-PAC® vs RAN-BMT) = 5.00. Hp-PAC® = lansoprazole-
of PPI-CM, PPI-CA and Hp-PAC while holding
(
clarithromycin-amoxicillin, prepackaged; MOH+ = Ontario Minis-
try of Health perspective excluding the patient copayment;
MOH– = MOH perspective including the patient copayment; PPI
the patient copayment constant at zero (i.e. MOH+
perspective). As anticipated, the expected costs for
all regimens increases linearly with the dispensing
fee, but the cost difference between PPI-CM and
PPI-CA is invariant to the increase in dispensing
fee. A threshold dispensing fee of $Can4.92 would
=
proton pump inhibitor; PPI-AM = PPI-amoxicillin-metronida-
zole; PPI-CA = PPI-clarithromycin-amoxicillin; PPI-CM = PPI-clar-
ithromycin-metronidazole; RAN-BMT = ranitidine-bismuth-
metronidazole-tetracycline; $Can = Canadian dollars.It should be
noted that the impact of metronidazole and/or clarithromycin
resistance on eradication rates is difficult to estimate. A contrib-
uting issue is the fact that a proportion of patients with resistant
organisms were probably included in the clinical trials from
which our base-case eradication rates are derived.
®
make the expected costs of Hp-PAC and PPI-CM
equal. Even at a dispensing fee of zero, the PPI-CA
®
regimen is more costly than Hp-PAC .
Discussion
tion rate of PPI-CM decreases, the expected costs
of PPI-CM increase. Therefore, Hp-PAC becomes
®
In this analysis, we have built upon the earlier
attractive relative to PPI-CM with respect to costs.
If the eradication rate of PPI-CM is lower than 88%
[4]
work by O’Brien et al. to compare the new Hp-
®
PAC with existing PPI-based regimens and RAN-
(
threshold value), Hp-PAC® also becomes both
BMT. Two perspectives were examined: (i) a strict
healthcare payer perspective excluding the patient
copayment (MOH+); and (ii) a healthcare payer
perspective including the patient copayment (MOH–).
more effective and less costly than PPI-CM.
In table V, we test the sensitivity of the results
to the individual PPI costs (in all areas of the model
©
Adis International Limited. All rights reserved.
Pharmacoeconomics 2001; 19 (8)

Prepackaged Regimen for H. pylori Eradication
841
Table V. Sensitivity analyses on expected 1-year cost per patient by alternative PPI prices, MOH+ perspective
Regimena
Expected cost per patient for 1 year ($Can)
base case (average PPI price) omeprazole price
lansoprazole price
pantoprazole price
RAN-BMT
PPI-CM
Hp-PAC®
PPI-CA
194
209
211
221
234
194
212
212
225
237
194
209
211
220
233
194
207
210
218
231
PPI-AM
®
a
Lansoprazole remains the PPI in the Hp-PAC . The PPI changes in the PPI-AM regimen, PPI-CA regimen and the PPI-CM regimen.
The PPI also changes in the PPI switch therapies and in the PPI maintenance therapy for all branches (HP1-HP5).
Hp-PAC® = lansoprazole-clarithromycin-amoxicillin, prepackaged; MOH+ = Ontario Ministry of Health perspective excluding the patient
copayment; PPI = proton pump inhibitor; PPI-AM = PPI-amoxicillin-metronidazole; PPI-CA = PPI-clarithromycin-amoxicillin; PPI-CM =
PPI-clarithromycin-metronidazole; RAN-BMT = ranitidine-bismuth-metronidazole-tetracycline; $Can = Canadian dollars.
The packaging of 3 separate chemical entities in
if this difference is robust. For example, from an
MOH+ perspective, when omeprazole is used as
the PPI in all PPI-based regimens (except Hp-
®
Hp-PAC under one Drug Identification Number
presents a unique opportunity to assess the impact
of drug packaging (and dispensing) on the cost ef-
fectiveness of duodenal ulcer therapy.
®
PAC ), the expected costs for PPI-CM and the Hp-
®
PAC are identical. In addition, if the PPI-CM
From an MOH+ perspective, cost-effectiveness
results indicate that RAN-BMT is the least costly
of the 5 regimens but also has the lowest effective-
ness (49 weeks per year without ulcer versus 51
weeks per year with the PPI-based regimens).
There was very little difference in the expected
number of weeks without ulcer between the PPI-
based regimens. Amongst these regimens, PPI-
CM is the least costly strategy and yields a cost-
effectiveness ratio of $Can5.77 compared with
RAN-BMT.
eradication is lowered from 90 to 88% (i.e. to equal
®
®
the eradication of Hp-PAC ), Hp-PAC becomes
the least costly strategy. Finally, our model did not
adjust for potentially enhanced compliance with
®
the Hp-PAC .
From our sensitivity analyses, the model is sen-
sitive to changes in the PPI-AM eradication rate.
However, PPI-AM is still dominated despite an in-
crease in eradication from 71 to 77%.
Metronidazole and/or clarithromycin resistance
rates are important factors that can alter the H. py-
lori eradication rate of a given strategy. To properly
address resistance, both the proportion of patients
with the resistant organism and the eradication rate
given resistance are needed. As we have shown,
there is a direct relationship between eradication
rate and expected costs and effects. However, the
true impact of antibacterial resistance on H. pylori
eradication rates is less clear. In addition, we do not
know the proportion of patients with resistant or-
ganisms that are included in the base case eradica-
tion rate assumptions.
From an MOH– perspective, RAN-BMT re-
mains the least costly regimen but the rank order-
®
®
ing of Hp-PAC and PPI-CM change. Hp-PAC
yields a cost-effectiveness ratio of $Can5 per ulcer
week averted compared with RAN-BMT.
The incremental cost effectiveness of Hp-PAC®
and PPI-CM largely depends on the dispensing
fees. The $Can2 difference in 1-year expected
®
costs between PPI-CM and Hp-PAC hinges on the
inclusion of the $Can2 copayment.
Regardless of the perspective, there is only an
annual $Can2 cost difference per patient between
PPI-CM and Hp-PAC® which, coincidentally, is
similar to the $Can2 dollar copayment in Ontario.
Given the various uncertainties and assumptions
made in the decision model, it is difficult to assess
There are a number of limitations to our analysis
including: (i) use of confirmed duodenal ulcer pa-
tients (rather than a dyspeptic patient population);
(ii) the assumption that the H. pylori eradication
rate for Hp-PAC® was equal to that of PPI-CA
©
Adis International Limited. All rights reserved.
Pharmacoeconomics 2001; 19 (8)

8
42
Agro et al.
2
2
2
70
50
30
PPI-CM
PPI-CA
Hp-PAC
2
1
10
90
1
1
70
50
0
2.11
4.11
6.11
8.11
10.11
12.11
14.11
Dispensing fee ($Can)
®
Fig. 3. Sensitivity analysis on the effect of dispensing fees on expected costs of PPI-CM, PPI-CA and Hp-PAC (MOH+ perspective;
patient copayment held constant at 0). The broken line indicates the threshold dispensing fee of $Can4.92 that would equalise the
expected costs of Hp-PAC and PPI-CM. Hp-PAC® = lansoprazole-clarithromycin-amoxicillin, prepackaged; MOH+ = Ontario Ministry
of Health perspective excluding the patient copayment; PPI = proton pump inhibitor; PPI-CA = PPI-clarithromycin-amoxicillin;
PPI-CM = PPI-clarithromycin-metronidazole; $Can = Canadian dollars.
®
®
(
which may not be justified if the Hp-PAC is as-
Acknowledgements
sociated with greater compliance and hence greater
efficacy); (iii) limited generalisability beyond On-
tario, Canada; (iv) not addressing antibacterial re-
sistance; and (v) not examining the impact of Hp-
Research assistance was provided by Jan Watson, Chris-
tine Henderson and Renee Crowther, MSc. This project was
supported by a grant from Abbott Laboratories Limited. Dr
O’Brien is a recipient of a career award in health sciences
from the Medical Research Council and Prescription Drug
Manufacturers Association of Canada.
®
PAC on patient compliance. Both of these last 2
®
factors are likely to be in favour of the Hp-PAC
and therefore our current analysis might be consid-
®
ered conservative with respect to Hp-PAC . To ad-
References
dress such limitations, further clinical and eco-
nomic research in this area is necessary.
1. NIH Consensus Development Panel. Helicobacter pylori in
peptic ulcer disease. JAMA 1994; 272: 65-9
2
. Hunt R, Thomson AB. Canadian Helicobacter pylori consensus
conference. Canadian Association of Gastroenterology. Can J
Gastroenterol 1998; 12 (1): 31-41
Conclusion
3. O’Brien BJ, Goeree R, Mohamed AH, et al. Cost-effectiveness
of Helicobacter pylori eradication for the long-term manage-
ment of duodenal ulcer in Canada. Arch Intern Med 1995;
Using the example of H. pylori eradication for
ulcers, we have shown that, in jurisdictions where
pharmacists are reimbursed for each drug item
within a dispensed prescription, the introduction of
pre-packaged multidrug regimens has the potential
to bring cost savings to both insurers (public or
private) and patients. In the specific comparison of
1
55: 1958-64
4. O’Brien BJ, Goeree R, Hunt R, et al. Economic evaluation of
alternative therapies in the long term management of peptic
ulcer disease and gastroesophageal reflux disease. Ottawa:
CCOHTA; 1996. Report no.: RFP#9503-OMEP: 1-130
5
. O’Brien BJ, Goeree R, Hunt R, et al. Cost effectiveness of al-
ternative Helicobacter pylori eradication strategies in the
management of duodenal ulcer. Can J Gastroenterol 1997; 11
(
4): 323-31
6
. Taylor JL, Zagari M, Murphy K, et al. Pharmacoeconomic com-
parison of treatments for the eradication of Helicobacter py-
lori. Arch Intern Med 1997; 157: 87-97
®
Hp-PAC versus PPI-CM, incremental costs are
sensitive to the level of dispensing fee and patient
copayment.
7. Sonnenberg FA, Beck JR. Markov models in medical decision
making. Med Decis Making 1993; 13: 322-38
©
Adis International Limited. All rights reserved.
Pharmacoeconomics 2001; 19 (8)

Prepackaged Regimen for H. pylori Eradication
843
8
. Veldhuyzen van Zanten SJ, Sherman PM, Hunt RH. Helicobac-
ter pylori: new developments and treatments. CMAJ 1997;
15. Park RE, Fink A, Brook RH. Physician ratings of appropriate
indications for six medical and surgical procedures. Am J
Public Health 1986; 76: 766-72
16. Ontario Ministry of Health. Schedule of benefits: physician ser-
vices under the Health Insurance Act. Ontario: Ministry of
Health, 1992
17. Gold MR, Siegel JE, Russell LB, et al. Cost-effectiveness in
health and medicine. New York (NY): Oxford University
Press, 1996
18. Holbrook AM, for the Ontario GI Therapy Review Panel. Med-
ical treatment guidelines for peptic ulcer disease and gas-
troesophageal reflux disease. Toronto: Ministry of Health,
1999. In press
1
56 (11): 1565-74
9
. DerSimonian R, Laird NM. Meta-analysis in clinical trials.
Control Clin Trials 1986; 7: 177-8
1
0. Fliess JL. The statistical basis of meta-analysis. Stat Methods
Med Res 1993; 2: 121-45
11. Biggerstaff BJ, Tweedie RL. Incorporating variability in esti-
mates of heterogeneity in the random effects model in meta
analysis. Stat Med 1997; 16: 753-68
1
2. Huang J-Q, Chiba N, Wilkinson JM, et al. Which combination
therapy can eradicate ≥90% Helicobacter pylori (H. pylori)
infection? Ameta-analysis of amoxicillin, metronidazole, tet-
racycline and clarithromycin containing regimens [abstract].
Gastroenterology 1997; 112 (4): A19
1
1
3. Graham DY, Lew GM, Malaty HM, et al. Factors influencing
the eradication of Helicobacter pylori with triple therapy.
Gastroenterology 1992; 102: 493-6
4. Penston JG. Review article – Helicobacter pylori: under-
standable caution but no excuse for inertia. Aliment Phar-
macol Ther 1994; 8: 369-89
Correspondence and offprints: Karen Agro, BScPhm,
PharmD, Centre for Evaluation of Medicines, St Joseph’s
Hospital, 105Main Street East, LevelP1, Hamilton, Ontario,
Canada L8N 1G6.
©
Adis International Limited. All rights reserved.
Pharmacoeconomics 2001; 19 (8)