January 2018
assert our reported time trends of Clostridium difficile
infection (CDI) incidence may be inaccurate because of
change in the method of testing. However, the method of
testing changed in only the last year of the study and
excluding that year in a sensitivity analysis had no effect on
the study findings. Second, they assert we do not provide
time of CDI occurrence. We do provide the timing of when
CDI was detected (sample collection date) and used previ-
ously described definitions of site of acquiring CDI (Supple-
mentary Table 1), which states that positive samples
collected within 48 hours of hospital admission be assigned
to the community-acquired group in the absence of other
recent admissions. We do not understand the basis of the
assertion that providers keep repeating CDI testing when
clinical suspicion is low. Third, they overlook that we have
indeed studied the effect of immunomodulators and anti-
tumor necrosis factor therapy on risk of CDI in a multivar-
iate analysis (Table 3). Fourth, although we did not have
direct markers of severity of IBD, we included several indi-
rect markers such as IBD surgery, hospitalization, mean
number of health care visits to providers, use of corticoste-
roids, use of immunomodulators, and use of biological ther-
apy. Last, in response to their suggestion that the data in our
electronic databases might have been cut, we would like to
point that we used a previously validated IBD dataset
(UMIBDED)1 and have also have validated the CDI dataset
used against a laboratory dataset and, moreover, found it to
be much more accurate than when the CDI codes in hospital
discharge summaries are used.2 The UMIBDED dataset was
the source of the cases with IBD and controls without IBD
selected only on the basis that they were matched according
to age, sex, and area of residence with IBD cases on the IBD
diagnosis date. All subjects in the UMIBDED who were still
alive and residing in Manitoba in the relevant time frame
were included. The CDI data included all positive test results
for everyone in the UMIBDED and was only cut to exclude
retesting within 14 days of a positive result, because these
tests are considered confirmatory rather than recurrent.
Although we agree on the need for prospective datasets, we
would point out the CDI dataset used in our study is a pro-
spectively collected dataset of laboratory-confirmed CDI
cases in the entire province of Manitoba, where CDI testing is
performed in laboratories only for individuals with diarrhea.
HARMINDER SINGH
Department of Internal Medicine
University of Manitoba and
University of Manitoba IBD Clinical and Research Center and
Community Health Sciences
University of Manitoba
Winnipeg, Manitoba, Canada
ZOANN NUGENT
University of Manitoba IBD Clinical and Research Center and
Department of Epidemiology and Cancer Registry
CancerCare Manitoba
Winnipeg, Manitoba, Canada
LAURA E. TARGOWNIK
Department of Internal Medicine
University of Manitoba and
Correspondence 257
University of Manitoba IBD Clinical and Research Center
Winnipeg, Manitoba, Canada
CHARLES N. BERNSTEIN
Department of Internal Medicine
University of Manitoba and
University of Manitoba IBD Clinical and Research Center
Winnipeg, Manitoba, Canada
References
1. Bernstein CN, et al. Am J Epidemiol 1999;149:916–924.
2. Singh H, et al. PLoS One 2017;12(2):e0171266.
Conflicts of interest
The authors disclose the following: Dr Bernstein is supported in part by the
Bingham Chair in Gastroenterology. Dr Lix is supported by a Research
Manitoba Chair. Dr Singh has been on advisory board of Pendopharm and
has received research funding from Merck Canada. Dr Bernstein has served
on advisory boards for Abbvie Canada, Ferring Canada, Janssen Canada,
Shire Canada, Pfizer Canada and Takeda Canada. He has consulted to
Mylan Pharmaceuticals and Bristol Myers Squibb. He has received
unrestricted educational grants from Abbvie Canada, Janssen Canada, Shire
Canada, and Takeda Canada. He has been on speaker’s bureau for Abbvie
Canada, Ferring Canada, and Shire Canada.
Most current article
https://doi.org/10.1053/j.gastro.2017.11.027
Prevalence of Helicobacter
pylori Infection in Asia:
Remembrance of Things Past?
Dear Editors:
We read the systematic review and meta-analysis by
Hooi et al1 with great interest. Their efforts should be
commended for updating and expanding on global preva-
lence data on Helicobacter pylori infection that were
published a few years ago.2 However, the data presented in
their article did not capture current epidemiologic changes
and the status quo in Asian countries, particularly those in
Japan. Our estimate3 finds the prevalence of H pylori
in Japan is rapidly decreasing owing in part to the abrupt
increase in the number of patients who have been cured of
H pylori infection (around 1.4 million per year from around
0.6 million from previous years) since the 2013 change in
the health insurance policy to cover H pylori gastritis. Our
prediction model shows current prevalence of the infection
in Japan was estimated at around 27% in 2016.3 Although
the prevalence of H pylori in Japan has been decreasing
overall because of the decreasing elderly population with
a high rate of infection (accounting for 11.6 million of
the decrease) and the emergence of new generations with
very low infection rates (cohort phenomenon), the
change in Japan’s insurance policy has accelerated the
decrease. Indeed, it facilitated additional 13.6 million
eradication cases over the last 16 years. Thus, these 2
factors—changes in the cohort and increased number of
eradications—contributed >20% of the decrease from the
258 Correspondence
estimated prevalence from 2000. Meta-analytic methodol-
ogy relying on previously published data cannot accurately
reflect such rapid change occurring in Japan. Hooi et al
lumped together 11 references published from 1993 to
2014 from which they computed Japan’s prevalence as
51.7% (95% CI, 44.7%–58.7%). Even with their subgroup
analysis based on more recent data (2000–2014), the
number was still much higher (46.5%; 95% CI, 36.5%–
56.4%) than our estimate of 27%, which happens to be
close to the 27.5% of the prevalence in a more recent data4
adopted in their paper. One of the reasons for the
discrepancy is that actual samplings in the majority of
these Japanese papers took place long before the eradica-
tion therapy commenced. Another reason their estimated
prevalence was conspicuously higher than ours may be that
their data were not representative of the general popula-
tion. Most of the sera were obtained from a population
above a certain age, group when health screenings were
provided; therefore, a limited number of data for younger
people (<20 years of age) were included.
We are also concerned that some of their data on Asian
countries were not instructive for a realistic understanding
of the prevalence of H pylori infection. For instance,
prevalence data were not shown for countries such as
Indonesia and Myanmar. However, those data for such
countries do exist, but were excluded because of differ-
ences in the target population.5,6 However, these data are
useful for estimating variations in the countries’ infection
rates. In fact, Indonesian’s data showing different infection
rates among the major islands that are geographically far
apart and consist of diverse ethnic groups are more
instructive than estimates for a single island. In a similar
vein, a recent report on Thailand’s infection rate, which
included greater areas of that country, demonstrated
highly variable prevalence among the different regions
(from 14.4% in the south to 60.6% in the northeast)7 as
compared with the single point data of 43.6% from the
country8 used for their article.
As the authors stated, the data in their study had a number
of limitations and, therefore, should be interpreted with
caution. The current prevalence data with broader repre-
sentation of general populations are evidently more useful
for strategic planning and provision for health policies on
H pylori infection in specific countries and/or regions.
KENTARO SUGANO
Department of Medicine
Jichi Medical University
Shimotsuke, Tochigi, Japan
SHINZO HIROI
Japan Medical Affairs
Takeda Pharmaceutical Company Limited
Chuo-ku, Tokyo, Japan
YOSHIO YAMAOKA
Department of Environmental and Preventive Medicine
Faculty of Medicine
Oita University
Yufu, Oita, Japan
Gastroenterology Vol. 154, No. 1
References
1. Hooi JKY, et al. Gastroenterology 2017;153:420–429.
2. Peleteiro B, et al. Dig Dis Sci 2014;59:1698–1709.
3. Hiroi S, et al. BMJ Open 2017;7:e015855.
4. Hirayama Y, et al. J Gastroenterol Hepatol 2014;
29(Suppl 4):16–19.
5. Syam AF, et al. PLoS One 2015;10(11):e0140186.
6. Myint T, et al. World J Gastroenterol 2015;21:629–636.
7. Uchida T, et al. PLoS One 2015;10(9):e0136775.
8. Hirai I, et al. Microbes and Infection 2010;12:227–230.
Conflicts of interest
The authors have made the following disclosures: KS had received research
grant from Takeda Pharmaceuticals. He also received lecture fees from
Takeda, Astellas, Eisai, Biofermin, and Biocodex. SH is an employee of
Takeda Pharmaceutical. YY declares no conflicts.
Most current article
https://doi.org/10.1053/j.gastro.2017.08.074
Reply. We appreciate your insightful comments
and thank you for sharing your recent publication
in BMJ Open. This is an important addition to the
literature, which demonstrated a decreasing trend in the
prevalence of Helicobacter pylori in Japan in more recent
years.1 In our systematic review, because of the inclusion
of a large number of papers we have divided them into 2
time periods (1970-1999 and 2000-2016), which provided
us with approximately one-half of all the papers in each
time frame. It was challenging to provide more granular
data by pooling prevalence over multiple shorter time
periods owing to an insufficient number of publications for
all the included countries.2 Because we have adopted a
uniform method to analyze and present the time trend data
for all countries, we understand that this approach will
inevitably limit detailed information on epidemiologic
changes for specific countries such as Japan. Given that
pooled H pylori prevalence was calculated from 2000 to
2016, such data are unlikely to reflect a decrease in
H pylori prevalence during the 16 years when a more
significant decrease in prevalence had occurred in the later
time period. As the authors nicely highlighted, H pylori
eradication occurred in Japan in 2000 and samplings in the
majority of studies from Japan probably occurred long
before the eradication therapy was commenced. We
recognized the limitations and have included an updated
descriptive details of more recent changes in the preva-
lence of H pylori in an interactive choropleth map (https://
people.ucalgary.ca/wggkaplan/HP2016.html). In addition,
we have excluded the study from Myanmar because the
target population had dyspeptic symptoms and may be
more likely to have a higher prevalence of H pylori, which
could bias the results. Nonetheless, we agree with the
authors that studies of individual countries, particularly
that of newly industrialized countries, remain crucial for
estimating variations in infection rates within an individual
country or across a specific region for better health care
provision on H pylori infection.