See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/330781139

Proton pump inhibitors therapy and the risk of pneumonia: a systematic

review and meta‐analysis of randomized controlled trials and observational
studies

Article  in  Expert Opinion on Drug Safety · January 2019

DOI: 10.1080/14740338.2019.1577820

CITATIONS READS

0 39

9 authors, including:

Chien-Chang Lee
National Taiwan University Hospital
206 PUBLICATIONS   3,175 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Use of antipsychotics increases the risk of fracture View project

All content following this page was uploaded by Wan-Ting Hsu on 16 March 2019.

The user has requested enhancement of the downloaded file.

 Expert Opinion on Drug Safety

ISSN: 1474-0338 (Print) 1744-764X (Online) Journal homepage: https://www.tandfonline.com/loi/ieds20

Proton pump inhibitors therapy and the

risk of pneumonia: a systematic review and
meta‐analysis of randomized controlled trials and
observational studies

Chih-Hung Wang, Cheng-Han Li, Ronan Hsieh, Cheng-Yi Fan, Tze-Chun Hsu,
Wei-Che Chang, Wan-Ting Hsu, Yu-Ya Lin & Chien-Chang Lee

To cite this article: Chih-Hung Wang, Cheng-Han Li, Ronan Hsieh, Cheng-Yi Fan, Tze-
Chun Hsu, Wei-Che Chang, Wan-Ting Hsu, Yu-Ya Lin & Chien-Chang Lee (2019): Proton
pump inhibitors therapy and the risk of pneumonia: a systematic review and meta‐analysis of
randomized controlled trials and observational studies, Expert Opinion on Drug Safety, DOI:
10.1080/14740338.2019.1577820

To link to this article: https://doi.org/10.1080/14740338.2019.1577820

View supplementary material

Accepted author version posted online: 31

Jan 2019.
Published online: 22 Feb 2019.

Submit your article to this journal

Article views: 43

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=ieds20
EXPERT OPINION ON DRUG SAFETY
https://doi.org/10.1080/14740338.2019.1577820

ORIGINAL RESEARCH

Proton pump inhibitors therapy and the risk of pneumonia: a systematic review and
meta-analysis of randomized controlled trials and observational studies
Chih-Hung Wang*a,b, Cheng-Han Li*c, Ronan Hsiehd, Cheng-Yi Fanc, Tze-Chun Hsua, Wei-Che Changc, Wan-Ting Hsue,
Yu-Ya Linf and Chien-Chang Lee a,b,e
a
Department of Emergency Medicine, National Taiwan University Hospital, Taipei, Taiwan; bDepartment of Emergency Medicine, College of
Medicine, National Taiwan University, Taipei, Taiwan; cCollege of Medicine, National Taiwan University, Taipei, Taiwan; dDepartment of Medicine,
Albert Einstein Medical Center, Philadelphia, PA, USA; eDepartment of Epidemiology, Harvard School of Public Health, Boston, MA, USA;
f
Department of Pharmacy, E-Da hospital, Kaohsiung, Taiwan

ABSTRACT ARTICLE HISTORY

Objective: We aimed to summarize the current evidence regarding the risk of pneumonia associated Received 10 November 2018
with proton pump inhibitors (PPI) treatment. Accepted 30 January 2019
Methods: We searched PubMed, Embase, and CENTRAL from the 1970 through December 2017. We KEYWORDS
included both randomized controlled trials (RCTs) and observational studies. We used random-effect Meta-analysis; pneumonia;
model to calculate the summary effect estimates and quantified the heterogeneity by I2 statistics. proton pump inhibitors;
Results: A total of 7,643,982 patients from 10 RCTs and 48 observational studies were included in this meta- protopathic bias
analysis. The primary meta-analysis demonstrated PPIs use was significantly associated with increased risk of
pneumonia, but the heterogeneity was high (odds ratio [OR], 1.43; 95% confidence interval [CI], 1.30–1.57; I2,
95.4%). The sensitivity analysis indicated PPIs were not statistically associated with increased risk of pneumonia
among patients concomitantly taking nonsteroidal anti-inflammatory drugs (OR, 1.11; 95% CI, 0.94–1.31; I2,
5.8%). The funnel plot demonstrated significant publication bias, especially for observational studies.
Conclusions: The presence of significant between-study heterogeneity and publication bias raised
concerns regarding the validity of the primary meta-analytic result. Protopathic bias, or reverse caus-
ality, may cause overestimated association. Studies that adopted a design to account for protopathic
bias did not show a significant association between PPI use and risk of pneumonia.

1. Introduction Nevertheless, most studies included in these meta-analyses

[4–6] were observational in design and inherently susceptible
Proton pump inhibitors (PPIs) have been widely used for acid-
to several biases, such as protopathic bias or confounding
related gastrointestinal disorders. PPI therapy has been associated
bias. Protopathic bias, or reverse causality, may have been
with several adverse effects [1]; among these, the association
present when a particular therapy was started due to the
between PPI use and an increased risk of pneumonia has recently
early manifestation of an undiagnosed disease [7]. Studies
attracted considerable attention.
have indicated that when PPIs are used even for a period as
In a large-scale case-control study published in 2004, Laheij
short as 7 days, there is an approximate four-fold increased
et al. [2], first indicated that among patients currently using PPIs,
risk of pneumonia [5]. Biologically, it is not plausible that PPIs
the adjusted relative risk of pneumonia was 1.89 (95% confi-
could increase the risk of pneumonia within such a short
dence interval [CI], 1.36–2.62) in comparison with those who
period of time. An alternative explanation for this observed
did not use PPIs. It is hypothesized that the elevation of gastric
association is the presence of protopathic bias, i.e. PPIs were
pH caused by PPIs leads to overgrowth of bacteria in the upper
prescribed for the early symptoms of pneumonia, such as non-
gastrointestinal tract, which may subsequently move upwards
specific epigastric or chest symptoms, which were mistakenly
and colonize the lower respiratory tract via gastroesophageal
treated as gastroesophageal reflux disease. Restricting ana-
reflux and microaspiration [3]. Following the study by Laheij
lyses to patients taking nonsteroidal anti-inflammatory drugs
et al. [2], many further research studies have been conducted
(NSAIDs) for ulcer prevention, rather than for symptoms
to address this possible adverse effect of PPI therapy. By pooling
potentially relevant to pneumonia, may be a solution to cir-
relevant evidence, three meta-analyses have consistently
cumvent protopathic bias [8].
demonstrated that PPI therapy may not only be associated
We therefore conducted a systematic review and meta-
with an increased risk of community-acquired pneumonia
analysis of randomized controlled trials (RCTs) and observa-
(CAP) but also hospital-acquired pneumonia (HAP) [4–6].
tional studies to determine the association between PPI

CONTACT Chien-Chang Lee, cclee100@gmail.com Health Data Science Research Group, Department of Emergency Medicine, National Taiwan University
Hospital, No. 7, Chung-Shan South Road, Taipei 100, Taiwan
*
These authors contributed equally to this work.
Supplemental data for this article can be accessed here.
© 2019 Informa UK Limited, trading as Taylor & Francis Group
2 C.-H. WANG ET AL.
therapy and the risk of pneumonia. Subsequently, we further
performed sensitivity and subgroup analyses in an attempt to
identify a potential source of heterogeneity in the interpreta-
tion of the meta-analysis results.
2. Methods
2.1. Data sources and search strategy
We performed the systematic review and meta-analysis in
accordance with the Preferred Reporting Items for Systematic
Reviews and Meta-Analyses (PRISMA) [9] and Meta-analysis of
Observational Studies in Epidemiology (MOOSE) [10]. We
searched PubMed, Embase, and CENTRAL for articles pub-
lished from inception in 1970 to December 2017. Our search
strings included designed keywords for two concepts: pneu-
monia and acid suppressants (Supplemental Table 1). We set
no restrictions on publication year or language. To ensure
completeness, we screened relevant review articles and
meta-analyses for references not captured by our search strat-
egy and we also screened relevant conference abstracts for
available data.
2.2. Study selection
We reviewed each study reporting an association between PPI
exposure and incidence of pneumonia. We included studies
reporting available data for the calculation of effect estimates
of pneumonia risk amongst patients receiving PPI therapy.
Studies reporting information on CAP or HAP were eligible.
Based on the published guidelines of the Cochrane
Collaboration [11], we included both RCTs and observational
studies to maximize inclusion of available studies, and thereby
allowing subsequent sensitivity and subgroup analyses. We
did not exclude studies based on age, comorbidities, or clinical
setting. We excluded case reports, case series, editorials, clin-
ical guidelines, and studies on non-human subjects. For stu-
dies with substantially overlapping cohorts, we only included
studies with the largest cohort. Discrepancies between
reviewers as to which articles merited inclusion were resolved
by a consensus meeting of three authors.
2.3. Data extraction and quality assessment
Two authors independently extracted qualitative and quanti-
tative data, and disagreements were resolved by the third
author. Data were extracted for study location, number of
participants, population characteristics, clinical setting, defini-
tion of PPI exposure, diagnosis of pneumonia, and adjusted
effect estimates. For studies not providing adjusted effect
estimates, we calculated the unadjusted odds ratio (OR)
according to the reported number of patients with or without
PPI therapy and the respective number of events of
pneumonia.
We credited studies as good quality when they met the
following criteria: (1) NSAID users: restricting analyses to
NSAID users; (2) Adjustment for smoking: adjusting smoking
in the multivariate analyses; (3) Full adjustment: comprehen-
sively adjusting potential confounders in the multivariate
analyses; (4) Pneumonia validation: validating the diagnosis
of pneumonia through radiographic or laboratory results; (5)
Different time periods: reducing the likelihood of exposure
misclassifications by categorizing drug-filling records into dif-
ferent time periods; (6) Sensitivity analysis: conducting sensi-
tivity analyses that took potential sources of bias into account.
We also used Scottish Intercollegiate Guideline Network (SIGN)
tool to evaluate each included study.
2.4. Statistical analyses
We used the metafor package in R 3.4.4. to conduct all statis-
tical analyses and generate plots. We used a random effects
model via the rma function to calculate the weighted mean,
variance of the summary effect, associated 95% CI, and
p-value. Since pneumonia was a rare clinical event, all mea-
sures of effect estimates were assumed to approximate the
odds ratio (OR) and were pooled in meta-analyses.
Heterogeneity was estimated using restricted maximum-
likelihood estimation [12] and quantitated using I2 statistics,
with I2 statistics > 50% deemed as the presence of significant
heterogeneity [13].
To assess potential sources of significant heterogeneity, we
performed several sensitivity and subgroup analyses. In sensi-
tivity analyses, we examined the robustness of the estimated
risk of pneumonia by restricting included studies to the fol-
lowing specific groups: (1) Comprehensive adjustment, includ-
ing studies adjusting potential confounders in the multivariate
analyses; (2) Frailty adjustment, including studies with effect
estimates adjusting for prior physician visits or hospitalization;
and (3) NSAID users, including studies restricting study cohort
to NSAID users. In subgroup analyses, we examined the PPI-
related risk of pneumonia stratified by the study design, study
location, number of participants, type of PPIs, type of pneu-
monia, cumulative duration of PPI therapy, and the interval
between initiation of PPIs and the diagnosis of pneumonia.
We examined the presence of publication bias via funnel
plots and Egger’s test.
3. Results
3.1. Search results and study characteristics
According to the search strategy, we identified 661 studies for
full-text review. According to the selection criteria, 65 studies
were included in the systematic review [2,8,14–76] (Figure 1).
As shown in Supplemental Table 2, there were 27 case–
control studies, 25 cohort studies, and 10 RCTs included in the
systematic review. Overall, 17 studies specifically examined the
association between PPI use and CAP, and 16 studies focused
on HAP. There were 37 studies that comprehensively adjusted
for potential confounding factors and 22 studies that con-
firmed the diagnosis of pneumonia through either radio-
graphic findings or laboratory results. The results of the
quality assessment are shown in Supplemental Figure 1.
Notably, only five studies reported PPI-related risks of pneu-
monia among NSAID users. As shown in Supplemental Table
3A-C, most of the included studies were judged to have high
or acceptable quality according to SIGN tool.

Figure 1. Flow diagram of study selection.
3.2. Meta-analyses
Seven studies [25,26,30,32,41,44,76] were not pooled in the
meta-analysis due to substantially overlapping populations
(Figure 1). The remaining 58 studies, including 10 RCTs and
48 observational studies, were synthesized in the meta-
analysis with a total of 7,643,982 patients (Figure 2; Table 1).
Overall, the primary meta-analysis shows that PPI use was
significantly associated with an increased risk of pneumonia,
but the heterogeneity was high (OR, 1.43; 95% CI, 1.30–1.57; I2,
95.4%; Figure 2; Table 1).
As shown in Table 1 and Supplemental Figure 2, five studies,
involving 4,259,677 patients, investigated the association
EXPERT OPINION ON DRUG SAFETY 3
between PPI use and risk of pneumonia, specifically for NSAIDs
users. When these five studies were pooled, the heterogeneity
decreased and the association between PPI use and risk of
pneumonia became statistically non-significant (OR, 1.11; 95%
CI, 0.94–1.31; I2, 5.8%). The subgroup analyses also indicated that
the heterogeneity decreased significantly when the studies were
pooled separately by study design, number of participants, type
of PPIs, and interval between the initiation of PPIs and diagnosis
of pneumonia. The 10 pooled RCTs demonstrated low hetero-
geneity and statistically non-significant results (OR, 1.13; 95% CI,
0.71–1.78; I2, 30.2%). In contrast, the pooled results for studies
with a patient number < 1000 indicated a statistically significant
association between PPI therapy and an increased risk of
4 C.-H. WANG ET AL.

Figure 2. Summary forest plot of overall risk of pneumonia associated with proton pump inhibitor use, subdivided by study design.
 EXPERT OPINION ON DRUG SAFETY 5

Table 1. Summary of sensitivity and subgroup analyses.

Subgroup # of studies Pooled effect estimate I2 P-value Reference
Primary meta-analysis
Overall pneumonia risk related to PPI 58 1.43 [1.30, 1.57] 95.4% < 0.01 [2,8,14–24,27–29,31,33–40,42,43,45–75]
Sensitivity analyses
Comprehensive adjustment 33 1.37 [1.26, 1.49] 93.5% < 0.01 [2,8,20,25,28,29,33–39,42,43,46–53,56–58,63,67,69,71,73–75]
Frailty adjustment 10 1.23 [1.09, 1.40] 92.5% < 0.01 [2,8,29,33,35,38,42,43,48,67]
NSAID users 5 1.11 [0.94, 1.31] 5.8% 0.22 [8,16–18,33]
Subgroup analyses
Study design
Observational studies 48 1.45 [1.32, 1.59] 96.1% < 0.01 [2,8,24,27–29,31,33–40,42,43,45–75]
RCT 10 1.13 [0.71, 1.78] 30.2% 0.61 [14–23]
Study locations
North America 18 1.26 [1.08, 1.47] 88.4% < 0.01 [15,19,23,28,31,35–38,42,46–48,52,63,64,66,75]
Europe 19 1.52 [1.29, 1.79] 97.8% < 0.01 [2,14,16,20,24,27,29,33,39,40,45,50,51,55,59,62,67,69,72]
Asia 14 1.54 [1.28, 1.83] 74.0% < 0.01 [17,18,21,43,53,54,56–58,65,68,71,73,74]
Other regions 6 1.42 [1.02, 1.98] 50.6% 0.04 [22,34,49,60,61,70]
Number of participants
< 1000 22 1.57 [1.18, 2.09] 46.1% < 0.01 [14,15,17–19,21–23,28,31,54,55,59–62,64–66,69–71]
≥ 1000 36 1.40 [1.28, 1.54] 96.5% < 0.01 [2,8,16,20,24,27,29,33–40,42,43,45–53,56–58,63,67,68,72–75]
Different PPIs
Omeprazole 4 1.58 [1.04, 2.41] 72.8% 0.03 [2,14,33,71]
Pantoprazole 5 1.82 [1.35, 2.45] 1.6% < 0.01 [2,22,23,33,66]
Lansoprazole 6 1.17 [1.00, 1.37] 0.0% 0.05 [2,17–19,21,33]
Esomeprazole 5 0.88 [0.54, 1.45] 26.0% 0.63 [2,15,16,20,33]
Different pneumonia type
CAP 17 1.44 [1.30, 1.59] 92.6% < 0.01 [2,8,24,25,27,29,33,35–37,39,42,46,51,63,67,70]
HAP 16 1.50 [1.15, 1.95] 86.5% < 0.01 [14,21–23,28,31,38,50,52,53,59,62,66,68,69,71]
Duration of PPI therapy
< 30 days 4 1.57 [1.19, 2.07] 61.4% < 0.01 [2,29,33,58]
≥ 30 days 6 1.22 [0.99, 1.50] 88.5% 0.06 [2,29,33,35,49,58]
Initiation of PPI therapy
Initiated in 7 days 3 3.50 [2.86, 4.28] 0.0% < 0.01 [25,29,49]
Initiated in 30 days 3 2.49 [2.10, 2.94] 0.0% < 0.01 [29,39,49]
CAP: community-acquired pneumonia; GERD: gastroesophageal reflux disease; HAP: hospital-acquired pneumonia; PPI: proton-pump inhibitors; RCT: randomized
controlled trial.

pneumonia (OR, 1.57; 95% CI, 1.18–2.09; I2, 46.1%). Moreover,
there was an increased risk of pneumonia associated with pan-
toprazole or lansoprazole therapy (pantoprazole: OR, 1.82; 95%
CI, 1.35–2.45; I2, 1.6%; lansoprazole: OR, 1.17; 95% CI, 1.00–1.37;
I2, 0.0%) and in patients receiving PPIs for less than 7 or 30 days
(7 days: OR, 3.50; 95% CI, 2.86–4.28; I2, 0.0%; 30 days: OR, 2.49;
95% CI, 2.10–2.94; I2, 0.0%).
The funnel plot that included all studies (n = 58) in the
primary meta-analysis demonstrated clustering of studies in
favor of a positive association (Figure 3(a)). The Egger’s test
confirmed asymmetry of the funnel plot (p-value, 0.01), indi-
cating the presence of publication bias. When publication bias
was examined according to the study design, only the funnel
plot that included observational studies was significantly
asymmetrical (p-value, 0.01; Figure 3(b)), while the funnel
plot that included RCTs was not (p-value, 0.30; Figure 3(c)).
4. Discussion
4.1. Main findings
The systematic review identified 65 studies examining the
relationship between PPI use and risk of pneumonia. The
primary meta-analysis of 58 studies, involving 7,643,982
patients, indicates that PPI use may be associated with an
increased risk of pneumonia (OR, 1.43; 95% CI, 1.30–1.57; I2,
95.4%). However, due to the great heterogeneity between
studies and the presence of publication bias, this conclusion
should be interpreted with caution. Through sensitivity and
subgroup analyses, several possible sources of heterogeneity
were identified, including patient population, study design,
and interval between the initiation of PPIs and diagnosis of
pneumonia, suggesting the possible presence of protopathic
bias in the meta-analyses of observational studies.
4.2. Comparisons with previous studies
Previous meta-analyses [4–6] have indicated the significant asso-
ciation between PPI use and an increased risk of pneumonia. It
has also been noted that the inter-study heterogeneity is high;
however, limited by the number of included studies, the poten-
tial sources of heterogeneity could not be identified. These
studies [4–6] demonstrated that newly prescribed PPI users
were associated with a higher risk of pneumonia in comparison
with chronic users. Similarly, our results indicate that even
a period as short as 7 days of PPI use was significantly associated
with an increased risk of pneumonia. Moreover, the interval
between the initiation of PPIs and diagnosis of pneumonia may
be a possible cause of significant heterogeneity.
At intragastric pH values less than 4.0, 99.9% of the bacteria
are killed within 30 minutes [77]. It is hypothesized that PPIs
may predispose individuals to pneumonia due to the fact that
elevated intragastric pH may contribute to bacterial over-
growth in the stomach [78], which subsequently increases
6 C.-H. WANG ET AL.

(a)

(b)

(c)

Figure 3. (a) Funnel plot of all studies included in the primary meta-analysis. (b) Funnel plot of observational studies. (c) Funnel plot of randomized controlled trials.

the risk of micro-aspiration of bacteria [79]. However, it has
been reported that following initiation of 30 mg day−1 lanso-
prazole, the median 24-hour intragastric pH value on days 1, 2
and 7 remained slightly below 4 [80]. Therefore, it is not
biologically plausible that pneumonia could occur in such
a short time frame.
Furthermore, tachyphylaxis has not been reported in PPI
use [81]; therefore, the suppression of gastric acid production
and the elevation of intragastric pH should be more pro-
nounced in patients receiving long-term PPIs. Nevertheless,
as shown in previous studies [4–6] and ours, long-term PPI
use was not or much less significantly associated with an
increased risk of pneumonia in comparison with short-term
use. Therefore, these results raise concerns about protopathic
bias in investigating the association between PPI use and risk
of pneumonia among observational studies.
4.3. Interpretation of sensitivity and subgroup analyses
Protopathic bias, or reverse causality, is a source of systematic
bias that occurs when exposure status may change in response
to the manifestation of a latent outcome. PPIs may be prescribed
for the early symptoms of pneumonia, such as non-specific chest
discomfort, and when pneumonia is later diagnosed, PPIs may
fallaciously appear to cause pneumonia. Protopathic bias has
been noted in research investigating antibiotics and antibiotic-
resistant infections [82], alcohol consumption and systemic lupus
[83], and NSAIDs and cancer [84]. Moreover, protopathic bias has
been noted in research regarding PPIs and gastric cancer [85] or
ischemic heart disease [86].
To deal with protopathic bias, a lag-time approach may be
a useful strategy in observational studies [85,87]. Alternatively,
restricting analysis to NSAID users may help to minimize pro-
topathic bias in the study of PPIs. For NSAID users, PPIs were
prescribed to prevent ulcers or dyspepsia rather than for the
early symptoms of pneumonia. As shown in our sensitivity
analyses of NSAID users, the pooled results demonstrate that
PPI use was not statistically associated with an increased risk
of pneumonia. In comparison with the primary meta-analysis,
the inter-study heterogeneity was much lower among the
studies of NSAID users. Therefore, by circumventing the pro-
blem induced by protopathic bias, the pooled results of stu-
dies on NSAID users appear to be less biased and more
credible.
In contrast, RCTs may be less susceptible to protopathic
bias, since RCTs adopted an intention-to-treat approach to
analyze the data, with PPI use started from the time of rando-
mization. Moreover, most of the included studies were obser-
vational in design, which could only establish an association
rather than a causal relationship between PPI use and risk of
pneumonia. One of the indications for prescribing PPIs is
gastroesophageal reflux disease [88], which by itself, may
contribute to an increased risk of pneumonia through micro-
aspiration [89]. Thus, confounding by indication in observa-
tional studies may lead to overestimated effect estimates,
which could only be avoided by RCTs. In our subgroup ana-
lyses, the pooled results of RCTs demonstrate no statistically
significant association between PPI use and risk of pneumonia
and low heterogeneity.
EXPERT OPINION ON DRUG SAFETY 7
However, in comparison with observational studies, the
number of included patients or follow-up duration in RCTs
may be lower or shorter than those in observational studies.
Additionally, in most included RCTs [15–20,22], the develop-
ment of pneumonia was not the primary outcome and was
not clearly defined. Therefore, observational studies using
healthcare databases could still provide valuable data for
researchers and regulatory agencies [90] to investigate the
safety of drugs, as long as the relevant bias could be well
controlled.
Subgroup analyses found that the heterogeneity may be
decreased if the studies were pooled according to different
types of PPIs. Nevertheless, the potency in suppressing gastric
acid secretion has not been shown to be significantly different
among different PPIs [91,92]. Therefore, whether different
types of PPIs could lead to different risks of pneumonia should
be examined further.
4.4. Study limitations
Important adverse effects may occur rarely, and accordingly,
systematic reviews and meta-analyses that pool harms data
from various sources can provide greater insight than a single
RCT or observational study. However, publication bias could
raise serious problems. The main study limitation of our meta-
analysis results was the presence of significant publication
bias, which seemed to be caused by the included observa-
tional studies. Meyer et al. [93], found evidence of publication
bias in gastroenterological research that abstracts with signifi-
cant findings were more likely to be published. More than half
of the abstracts presented from 2011 to 2013 in Americas
Hepato-Pancreato-Biliary Association Congresses failed to
reach publication [93]. Nevertheless, we searched relevant
conference abstracts and the publication bias remained
significant.
Publication bias [94] could jeopardize the validity of meta-
analyses. Studies [95,96] have demonstrated that once unpub-
lished data are pooled in the analysis, the conclusion may be
very different from the original results. Most studies [94] have
focused on the consequences of publication bias concerning
RCTs. In the present study, the most common study design
was observational; the effects of publication bias on this type
of study is unclear but may be important, since the incidence
of adverse events may be misestimated and inaccurate asso-
ciations between clinical variables may be assumed. This may
subsequently lead to withdrawals, revocations, or suspensions
of marketing authorizations. Post-marketing reporting of
adverse events during real-world use of a medicinal product
is important in identifying new, rare, or serious adverse reac-
tions with the potential to influence the benefit–risk balance
of a product [97,98]. Between 2012 and 2016, meta-analyses
were cited in 13 of the 18 regulatory actions within the EU
[99]; therefore, cautious inspection of available evidence con-
cerning the adverse events of a certain medication is
paramount.
Secondly, the results of subgroup analyses were based on
univariate analyses; accordingly, the results should be used for
exploratory purpose rather than viewed as definitive conclusions.
8 C.-H. WANG ET AL.
4.5. Applications in clinical practice
The meta-analytic results of sensitivity and subgroup analyses
suggested that the previously indicated associations [4–6]
between PPI therapy and increased risk of pneumonia may
be biased by the protopathic bias, publication bias and con-
founding by indication. According to the meta-analytic results
of RCTs and observational studies focusing on NSAID users,
PPI use may not be associated with increased risk of pneumo-
nia. Concern for pneumonia should not prevent clinicians from
initiating indicated PPI therapy if otherwise indicated [100].
5. Conclusion
Our primary meta-analysis demonstrates that PPI use may be
associated with an increased risk of pneumonia. Nevertheless,
the presence of significant between-study heterogeneity and
publication bias raised concerns regarding the validity of the
primary meta-analytic result. Protopathic bias, or reverse caus-
ality, may cause overestimated association. Studies that
adopted a design to account for protopathic bias did not
show a significant association between PPI use and risk of
pneumonia. Therefore, concern for pneumonia should not
prevent clinicians from initiating indicated PPI therapy if other-
wise indicated.
Acknowledgments
The authors would like to sincerely thank the staff of the Core Labs, the
Department of Medical Research, National Taiwan University Hospital for
technical assistance.
Author contributions
Cheng-Han Li and Cheng-Yi Fan involved in the data collection and extrac-
tion. Chih-Hung Wang and Cheng-Han Li performed the statistical analysis,
interpreted the results and wrote the draft. Tze-Chun Hsu and Wan-Ting Hsu
provided technical support of statistical analyses. Ronan Hsieh reviewed the
manuscript and provided critical insights. Wei-Che Chang and Yu-Ya Lin
helped with language editing and critically revised the manuscript. Chien-
Chang Lee obtained funding, conceive the idea, designed the study, and
guided the statistical analysis and manuscript writing, and supervise the
whole process. All authors approved the final version of the manuscript and
agree to be accountable for all aspects of the work.
Data availability statement
The data that support the findings of this study are available from the
corresponding author, upon reasonable request.
Funding
This manuscript was supported by the Taiwan National Ministry of Science
and Technology (MOST 104-2314-B-002-039-MY3) and National Taiwan
University Hospital (108-S4091).
Declaration of interest
The authors have no relevant affiliations or financial involvement with any
organization or entity with a financial interest in or financial conflict with
the subject matter or materials discussed in the manuscript. This includes
employment, consultancies, honoraria, stock ownership or options, expert
testimony, grants or patents received or pending, or royalties.
Reviewer disclosures
Peer reviewers on this manuscript have no relevant financial or other
relationships to disclose.
ORCID
Chien-Chang Lee http://orcid.org/0000-0002-1243-2463
References
Papers of special note have been highlighted as either of interest (•) or of
considerable interest (••) to readers.
1. Heidelbaugh JJ, Goldberg KL, Inadomi JM. Overutilization of proton
pump inhibitors: a review of cost-effectiveness and risk [corrected].
Am J Gastroenterol. 2009;104(Suppl 2):S27–32.
2. Laheij, R.J., et al. Risk of community-acquired pneumonia and use
of gastric acid-suppressive drugs. JAMA. 2004;292(16):1955–1960.
3. Sanduleanu, S., et al. Non-Helicobacter pylori bacterial flora during
acid-suppressive therapy: differential findings in gastric juice and
gastric mucosa. Aliment Pharmacol Ther. 2001;15(3):379–388.
4. Lambert, A.A., et al. Risk of community-acquired pneumonia with
outpatient proton-pump inhibitor therapy: a systematic review and
meta-analysis. PLoS One. 2015;10(6):e0128004.
• A previous systematic review and meta-analysis investigated
similar the same topic as the current work.
5. Eom, C.S., et al. Use of acid-suppressive drugs and risk of pneumo-
nia: a systematic review and meta-analysis. Cmaj. 2011;183
(3):310–319.
• A previous systematic review and meta-analysis investigated
similar the same topic as the current work.
6. Johnstone J, Nerenberg K, Loeb M. Meta-analysis: proton pump
inhibitor use and the risk of community-acquired pneumonia.
Aliment Pharmacol Ther. 2010;31(11):1165–1177.
• A previous systematic review and meta-analysis investigated
similar the same topic as the current work.
7. Horwitz RI, Feinstein AR. The problem of “protopathic bias” in case-
control studies. Am J Med. 1980;68(2):255–258.
8. Filion, K.B., et al. Proton pump inhibitors and the risk of hospitalisa-
tion for community-acquired pneumonia: replicated cohort studies
with meta-analysis. Gut. 2014;63(4):552–558.
9. Liberati, A., et al. The PRISMA statement for reporting systematic
reviews and meta-analyses of studies that evaluate health care
interventions: explanation and elaboration. Ann Intern Med.
2009;151(4):W65–94.
10. Stroup, D.F., et al. Meta-analysis of observational studies in epide-
miology: a proposal for reporting. Meta-analysis Of Observational
Studies in Epidemiology (MOOSE) group. JAMA. 2000;283
(15):2008–2012.
11. Anglemyer A, Horvath HT, Bero L. Healthcare outcomes assessed
with observational study designs compared with those assessed in
randomized trials. Cochrane Database Syst Rev. 2014;4:MR000034.
12. Viechtbauer W. Bias and efficiency of meta-analytic variance esti-
mators in the random-effects model. J Educ Behav Stat. 2005;30
(3):261–293.
13. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin
Trials. 1986;7(3):177–188.
14. Kantorova, I., et al. Stress ulcer prophylaxis in critically ill patients:
a randomized controlled trial. Hepatogastroenterology. 2004;51
(57):757–761.
15. American Lung Association Asthma Clinical Research, C, et al.
Efficacy of esomeprazole for treatment of poorly controlled
asthma. N Engl J Med. 2009;360(15):1487–1499.
16. Scheiman, J.M., et al. Prevention of peptic ulcers with esomepra-
zole in patients at risk of ulcer development treated with low-dose

acetylsalicylic acid: a randomised, controlled trial (OBERON). Heart.
2011;97(10):797–802.
17. Sugano, K., et al. Lansoprazole for secondary prevention of gastric
or duodenal ulcers associated with long-term low-dose aspirin
therapy: results of a prospective, multicenter, double-blind, rando-
mized, double-dummy, active-controlled trial. J Gastroenterol.
2011;46(6):724–735.
18. Sugano, K., et al. Lansoprazole for secondary prevention of gastric
or duodenal ulcers associated with long-term non-steroidal
anti-inflammatory drug (NSAID) therapy: results of a prospective,
multicenter, double-blind, randomized, double-dummy,
active-controlled trial. J Gastroenterol. 2012;47(5):540–552.
19. Lima, J.J., et al. Association of CYP2C19 polymorphisms and
lansoprazole-associated respiratory adverse effects in children.
J Pediatr. 2013;163(3):686–691.
20. Estborn L, Joelson S. Frequency and time to onset of
community-acquired respiratory tract infections in patients receiv-
ing esomeprazole: a retrospective analysis of patient-level data in
placebo-controlled studies. Aliment Pharmacol Ther. 2015;42
(5):607–613.
21. Lin, C.C., et al. Stress ulcer prophylaxis in patients being weaned
from the ventilator in a respiratory care center: a randomized con-
trol trial. J Formos Med Assoc. 2016;115(1):19–24.
22. Selvanderan, S.P., et al. Pantoprazole or Placebo for Stress Ulcer
Prophylaxis (POP-UP): randomized double-blind exploratory study.
Crit Care Med. 2016;44(10):1842–1850.
23. Alhazzani, W., et al. Withholding pantoprazole for stress ulcer
prophylaxis in critically Ill patients: a pilot randomized clinical trial
and meta-analysis. Crit Care Med. 2017;45(7):1121–1129.
24. van de Garde, E.M., et al. Angiotensin-converting enzyme inhibitor
use and pneumonia risk in a general population. Eur Respir J.
2006;27(6):1217–1222.
25. Gulmez, S.E., et al. Use of proton pump inhibitors and the risk of
community-acquired pneumonia: a population-based case-control
study. Arch Intern Med. 2007;167(9):950–955.
26. Hennessy, S., et al. Observed association between antidepressant
use and pneumonia risk was confounded by comorbidity
measures. J Clin Epidemiol. 2007;60(9):911–918.
27. Almirall, J., et al. New evidence of risk factors for
community-acquired pneumonia: a population-based study. Eur
Respir J. 2008;31(6):1274–1284.
28. Beaulieu, M., et al. Do proton-pump inhibitors increase the risk for
nosocomial pneumonia in a medical intensive care unit? J Crit Care.
2008;23(4):513–518.
29. Sarkar M, Hennessy S, Yang YX. Proton-pump inhibitor use and the
risk for community-acquired pneumonia. Ann Intern Med. 2008;149
(6):391–398.
30. Herzig, S.J., et al. Acid-suppressive medication use and the risk for
hospital-acquired pneumonia. JAMA. 2009;301(20):2120–2128.
31. Marciniak, C., et al. Examination of selected clinical factors and
medication use as risk factors for pneumonia during stroke reha-
bilitation: a case-control study. Am J Phys Med Rehabil. 2009;88
(1):30–38.
32. Myles, P.R., et al. Risk of community-acquired pneumonia and the
use of statins, ace inhibitors and gastric acid suppressants: a
population-based case-control study. Pharmacoepidemiol Drug
Saf. 2009;18(4):269–275.
33. Rodriguez, L.A., et al. Acid-suppressive drugs and
community-acquired pneumonia. Epidemiology. 2009;20
(6):800–806.
34. Roughead, E.E., et al. Proton-pump inhibitors and the risk of anti-
biotic use and hospitalisation for pneumonia. Med J Aust. 2009;190
(3):114–116.
35. Dublin, S., et al. Use of proton pump inhibitors and H2 blockers and
risk of pneumonia in older adults: a population-based case-control
study. Pharmacoepidemiol Drug Saf. 2010;19(8):792–802.
36. Eurich, D.T., et al. Recurrent community-acquired pneumonia in
patients starting acid-suppressing drugs. Am J Med. 2010;123
(1):47–53.
EXPERT OPINION ON DRUG SAFETY 9
37. Gau, J.T., et al. Pharmacotherapy and the risk for
community-acquired pneumonia. BMC Geriatr. 2010;10:45.
38. Redelmeier, D.A., et al. Postoperative pneumonia in elderly patients
receiving acid suppressants: a retrospective cohort analysis. BMJ.
2010;340:c2608.
39. Meijvis, S.C., et al. Microbial evaluation of proton-pump inhibitors
and the risk of pneumonia. Eur Respir J. 2011;38(5):1165–1172.
40. Pasina, L., et al. Prevalence and appropriateness of drug prescrip-
tions for peptic ulcer and gastro-esophageal reflux disease in
a cohort of hospitalized elderly. Eur J Intern Med. 2011;22
(2):205–210.
41. Ernst, P., et al. Pramipexole use and the risk of pneumonia. BMC
Neurol. 2012;12:113.
42. Hermos, J.A., et al. Risk of community-acquired pneumonia in
veteran patients to whom proton pump inhibitors were
dispensed. Clin Infect Dis. 2012;54(1):33–42.
43. Liu, C.L., et al. Angiotensin-converting enzyme inhibitor/angioten-
sin II receptor blockers and pneumonia risk among stroke patients.
J Hypertens. 2012;30(11):2223–2229.
44. Mullerova, H., et al. The natural history of community-acquired
pneumonia in COPD patients: a population database analysis.
Respir Med. 2012;106(8):1124–1133.
45. Nielsen, A.G., et al. The impact of statin use on pneumonia risk and
outcome: a combined population-based case-control and cohort
study. Crit Care. 2012;16(4):R122.
46. Jena AB, Sun E, Goldman DP. Confounding in the association of
proton pump inhibitor use with risk of community-acquired
pneumonia. J Gen Intern Med. 2013;28(2):223–230.
47. Juthani-Mehta, M., et al. Modifiable risk factors for pneumonia
requiring hospitalization of community-dwelling older adults: the
health, aging, and body composition study. J Am Geriatr Soc.
2013;61(7):1111–1118.
48. Long, M.D., et al. Increased risk of pneumonia among patients with
inflammatory bowel disease. Am J Gastroenterol. 2013;108
(2):240–248.
49. Ramsay, E.N., et al. Proton pump inhibitors and the risk of pneu-
monia: a comparison of cohort and self-controlled case series
designs. BMC Med Res Methodol. 2013;13:82.
50. Buendgens, L., et al. Administration of proton pump inhibitors in
critically ill medical patients is associated with increased risk of
developing Clostridium difficile-associated diarrhea. J Crit Care.
2014;29(4):696 e11–5.
51. de Groot, M.C., et al. Sources of heterogeneity in case-control
studies on associations between statins, ACE-inhibitors, and proton
pump inhibitors and risk of pneumonia. Eur J Epidemiol. 2014;29
(10):767–775.
52. Herzig, S.J., et al. Acid-suppressive medication use in acute stroke
and hospital-acquired pneumonia. Ann Neurol. 2014;76(5):712–718.
53. Ho, S.W., et al. Population-based cohort study on the risk of pneu-
monia in patients with non-traumatic intracranial haemorrhage
who use proton pump inhibitors. BMJ Open. 2014;4(11):e006710.
54. Im, J.P., et al. Proton pump inhibitor use before percutaneous
endoscopic gastrostomy is associated with adverse outcomes.
Gut Liver. 2014;8(3):248–253.
55. Kroupa, R., et al. Is there a real risk of the long term medical
treatment of gastroesophageal reflux disease? United European
Gastroenterol J. 2014;2(1):A264.
56. Chen, C.H., et al. Proton pump inhibitor usage and the associated
risk of pneumonia in patients with chronic kidney disease.
J Microbiol Immunol Infect. 2015;48(4):390–396.
57. Ho, S.W., et al. Risk of stroke-associated pneumonia with
acid-suppressive drugs: a population-based cohort study.
Medicine (Baltimore). 2015;94(29):e1227.
58. Lee, S.W., et al. Proton pump inhibitors did not increase risk of
pneumonia in patients with chronic obstructive pulmonary disease.
J Clin Med Res. 2015;7(11):880–883.
59. Merli, M., et al. The chronic use of beta-blockers and proton pump
inhibitors may affect the rate of bacterial infections in cirrhosis.
Liver Int. 2015;35(2):362–369.
10 C.-H. WANG ET AL.
60. Ozkan, M.C., et al. Use of proton pump inhibitors in hematopoietic
stem cell transplantation does not increase the frequency of febrile
neutropenia. Eur Rev Med Pharmacol Sci. 2015;19(21):4105–4110.
61. Terg, R., et al. Proton pump inhibitor therapy does not increase the
incidence of spontaneous bacterial peritonitis in cirrhosis:
a multicenter prospective study. J Hepatol. 2015;62(5):1056–1060.
62. Bonvicini, D., et al. Type of stress-ulcer prophylaxis and incidence of
ventilator-associated pneumonia. Intensive Care Med Exp. 2016;4.
63. Chatterjee, S., et al. Anticholinergic medication use and risk of
pneumonia in elderly adults: a nested case-control study. J Am
Geriatr Soc. 2016;64(2):394–400.
64. Courson, A.Y., et al. Routine prophylaxis with proton pump inhibi-
tors and post-transplant complications in kidney transplant recipi-
ents undergoing early corticosteroid withdrawal. Clin Transplant.
2016;30(6):694–702.
65. Kawanishi, K., et al. Risk factors for aspiration pneumonia after
endoscopic hemostasis. Dig Dis Sci. 2016;61(3):835–840.
66. McCaleb, R.V., et al. Clinical outcomes of acid suppressive therapy
use in hematology/oncology patients at an academic medical
center. Ann Pharmacother. 2016;50(7):541–547.
67. Othman F, Crooks CJ, Card TR. Community acquired pneumonia
incidence before and after proton pump inhibitor prescription:
population based study. BMJ. 2016;355:i5813.
•• This is an important cohort study using advanced statistical
approach to deal with possible confounding factors.
68. Sasabuchi, Y., et al. Risks and benefits of stress ulcer prophylaxis for
patients with severe sepsis. Crit Care Med. 2016;44(7):e464–9.
69. Six, S., et al. Hyperoxemia as a risk factor for ventilator-associated
pneumonia. Crit Care. 2016;20(1):195.
70. Vukadinović, D., et al. Factors associated with early treatment fail-
ure in adult hospitalized patients with community-acquired
pneumonia. Vojnosanit Pregl. 2017;74(9):803–813.
71. Arai, N., et al. Histamine H2-blocker and proton pump inhibitor use
and the risk of pneumonia in acute stroke: a retrospective analysis
on susceptible patients. PLoS One. 2017;12(1):e0169300.
72. Bang UC, Bendtsen F. The use of proton pump inhibitors is not
associated with an increased risk of infections in a cohort of
patients with alcoholic cirrhosis from danish nationwide registers.
Hepatology. 2017;66:286A.
73. Ho, S.W., et al. Association of proton pump inhibitors usage with
risk of pneumonia in dementia patients. J Am Geriatr Soc. 2017;65
(7):1441–1447.
74. Hsu, W.T., et al. Risk of pneumonia in patients with gastroesopha-
geal reflux disease: a population-based cohort study. PLoS One.
2017;12(8):e0183808.
75. Rane, P., et al. Risk of pneumonia associated with proton pump
inhibitor use among community dwelling adults with gastroeso-
phageal reflux disease. Value Health. 2017;20(5):A179–A180.
76. Zirk-Sadowski, J., et al. Risks of longer term proton pump inhibitor
exposure in 228,752 older adults. Eur Geriatric Med. 2017;8:S29–S30.
77. Giannella RA, Broitman SA, Zamcheck N. Gastric acid barrier to
ingested microorganisms in man: studies in vivo and in vitro. Gut.
1972;13(4):251–256.
78. Torres, A., et al. Stomach as a source of colonization of the respira-
tory tract during mechanical ventilation: association with
ventilator-associated pneumonia. Eur Respir J. 1996;9
(8):1729–1735.
79. Williams C, McColl KE. Review article: proton pump inhibitors and
bacterial overgrowth. Aliment Pharmacol Ther. 2006;23(1):3–10.
View publication stats
80. Harder, H., et al. Effect of 7-day therapy with different doses of the
proton pump inhibitor lansoprazole on the intragastric pH in
healthy human subjects. Scand J Gastroenterol. 1999;34
(6):551–561.
81. Tani N. [Application of proton pump inhibitor to special conditions;
Zollinger-Ellison syndrome]. Nihon Rinsho. 1992;50(1):154–159.
82. Schneider-Lindner, V., et al. Antibacterial drugs and the risk of
community-associated methicillin-resistant Staphylococcus aureus
in children. Arch Pediatr Adolesc Med. 2011;165(12):1107–1114.
83. Wang, J., et al. Alcohol consumption is not protective for systemic
lupus erythematosus. Ann Rheum Dis. 2009;68(3):345–348.
84. Gonzalez-Perez A, Garcia Rodriguez LA, Lopez-Ridaura R. Effects of
non-steroidal anti-inflammatory drugs on cancer sites other than
the colon and rectum: a meta-analysis. BMC Cancer. 2003;3:28.
85. Tamim H, Monfared AA, LeLorier J. Application of lag-time into
exposure definitions to control for protopathic bias.
Pharmacoepidemiol Drug Saf. 2007;16(3):250–258.
86. Ali AK. The link between proton pump inhibitors and ischemic
heart disease could be explained by protopathic bias:
a pharmacovigilance analysis. Value Health. 2015;18(7):A373–4.
87. Arfe A, Corrao G. The lag-time approach improved drug-outcome
association estimates in presence of protopathic bias. J Clin
Epidemiol. 2016;78:101–107.
88. Jacobson, B.C., et al. Who is using chronic acid suppression therapy
and why? Am J Gastroenterol. 2003;98(1):51–58.
89. Gaude GS. Pulmonary manifestations of gastroesophageal reflux
disease. Ann Thorac Med. 2009;4(3):115–123.
90. Rodriguez EM, Staffa JA, Graham DJ. The role of databases in drug
postmarketing surveillance. Pharmacoepidemiol Drug Saf. 2001;10
(5):407–410.
91. Calabrese, C., et al. Ninety-six-hour wireless oesophageal pH mon-
itoring following proton pump inhibitor administration in NERD
patients. Aliment Pharmacol Ther. 2008;28(2):250–255.
92. Hellstrom PM, Vitols S. The choice of proton pump inhibitor: does it
matter? Basic Clin Pharmacol Toxicol. 2004;94(3):106–111.
93. Meyer, C., et al. Is publication bias present in gastroenterological
research? An analysis of abstracts presented at an annual congress.
PeerJ. 2018;6:e4995.
94. Golder, S., et al. Reporting of adverse events in published and
unpublished studies of health care interventions: a systematic
review. PLoS Med. 2016;13(9):e1002127.
95. Doshi P, Jefferson T, Del Mar C. The imperative to share clinical
study reports: recommendations from the Tamiflu experience. PLoS
Med. 2012;9(4):e1001201.
96. Wieseler, B., et al. Access to regulatory data from the european
medicines agency: the times they are a-changing. Syst Rev.
2012;1:50.
97. Alves C, Macedo AF, Marques FB. Sources of information used by
regulatory agencies on the generation of drug safety alerts. Eur
J Clin Pharmacol. 2013;69(12):2083–2094.
98. Inacio P, Cavaco A, Airaksinen M. The value of patient reporting to
the pharmacovigilance system: a systematic review. Br J Clin
Pharmacol. 2017;83(2):227–246.
99. Lane S, Lynn E, Shakir S. Investigation assessing the publicly avail-
able evidence supporting postmarketing withdrawals, revocations
and suspensions of marketing authorisations in the EU since 2012.
BMJ Open. 2018;8(1):e019759.
100. Savarino V, Dulbecco P, Savarino E. Are proton pump inhibitors
really so dangerous? Dig Liver Dis. 2016;48(8):851–859.