|Home | About | Journals | Submit | Contact Us | Français|
Observational studies and randomized controlled trials have yielded inconsistent findings about the association between the use of acid-suppressive drugs and the risk of pneumonia. We performed a systematic review and meta-analysis to summarize this association.
We searched three electronic databases (MEDLINE [PubMed], Embase and the Cochrane Library) from inception to Aug. 28, 2009. Two evaluators independently extracted data. Because of heterogeneity, we used random-effects meta-analysis to obtain pooled estimates of effect.
We identified 31 studies: five case–control studies, three cohort studies and 23 randomized controlled trials. A meta-analysis of the eight observational studies showed that the overall risk of pneumonia was higher among people using proton pump inhibitors (adjusted odds ratio [OR] 1.27, 95% confidence interval [CI] 1.11–1.46, I2 90.5%) and histamine2 receptor antagonists (adjusted OR 1.22, 95% CI 1.09–1.36, I2 0.0%). In the randomized controlled trials, use of histamine2 receptor antagonists was associated with an elevated risk of hospital-acquired pneumonia (relative risk 1.22, 95% CI 1.01–1.48, I2 30.6%).
Use of a proton pump inhibitor or histamine2 receptor antagonist may be associated with an increased risk of both community- and hospital-acquired pneumonia. Given these potential adverse effects, clinicians should use caution in prescribing acid-suppressive drugs for patients at risk.
Recently, the medical literature has paid considerable attention to unrecognized adverse effects of commonly used medications and their potential public health impact.1 One group of medications in widespread use is acid-suppressive drugs, which represent the second leading category of medication worldwide, with sales totalling US$26.9 billion in 2005.2
Over the past 40 years, the development of potent acid-suppressive drugs, including proton pump inhibitors, has led to considerable improvements in the treatment of acid-related disorders of the upper gastrointestinal tract.3 Experts have generally viewed proton pump inhibitors as safe.4 However, potential complications such as gastrointestinal neoplasia, malabsorption of nutrients and increased susceptibility to infection have caused concern.5
Of special interest is the possibility that acid-suppressive drugs could increase susceptibility to respiratory infections because these drugs increase gastric pH, thus allowing bacterial colonization.6,7 Several previous studies have shown that treatment with acid-suppressive drugs might be associated with an increased risk of respiratory tract infections8 and community-acquired pneumonia in adults6,7 and children.9 However, the association between use of acid-suppressive drugs and risk of pneumonia has been inconsistent.10–13
Given the widespread use of proton pump inhibitors and histamine2 receptor antagonists, clarifying the potential impact of acid-suppressive therapy on the risk of pneumonia is of great importance to public health.14 Previous meta-analyses have focused on the role of acid-suppressive drugs in preventing stress ulcer,11,13,15 but none have examined pneumonia as the primary outcome.
The aim of this study was to summarize the association between the use of acid-suppressive drugs and the risk of pneumonia in observational studies and randomized controlled trials.
The procedures used for this meta-analysis were consistent with recent guidelines for reporting of meta-analyses. Specifically, we followed the Meta-analysis Of Observational Studies in Epidemiology (MOOSE) guidelines16 for observational studies and the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) statement17 for randomized controlled trials.
We searched for studies that reported an estimate of effect for a potential association between the use of acid-suppressive drugs and the risk of pneumonia. We included observational studies and randomized controlled trials that were published as original articles.
We searched MEDLINE (PubMed), Embase and the Cochrane Central Register of Controlled Trials (CENTRAL) in the Cochrane Library from inception to Aug. 28, 2009. We also searched the bibliographies of relevant articles to identify additional studies.
To identify observational studies, we used the following combinations of search terms: (“acid-suppressive therapy” OR “acid-suppressive drugs” OR “acid-suppressive medications” OR “gastric acid suppressants” OR “proton pump inhibitors” OR “proton pumps” OR omeprazole OR nexium OR lansoprazole OR rabeprazole OR pantoprazole OR esomeprazole OR “H2 receptor antagonists” OR “histamine2 receptor antagonists” OR cimetidine OR ranitidine OR famotidine OR nizatidine) AND (pneumonia OR “community-acquired pneumonia” OR “nosocomial pneumonia” OR “hospital-acquired pneumonia” OR “intensive care unit”). We restricted this search to studies involving humans that were published in English.
To identify randomized controlled trials, we used the following combinations of search terms: (“acid-suppressive therapy” OR “acid-suppressive drugs” OR “acid-suppressive medications” OR “gastric acid suppressants” OR “proton pump inhibitors” OR “proton pumps” OR omeprazole OR nexium OR lansoprazole OR rabeprazole OR pantoprazole OR esomeprazole OR “H2 receptor antagonists” OR “histamine2 receptor antagonists” OR cimetidine OR ranitidine OR famotidine OR nizatidine). We restricted this search to randomized controlled trials.
We included any study that met all of the following criteria: was a case–control study, cohort study or randomized controlled trial; investigated the association between use of acid-suppressive drugs and risk of pneumonia; quantified the outcome with adjusted odds ratios (ORs), relative risk or number of events, and corresponding 95% confidence intervals (CIs); and reported the results for proton pump inhibitors and histamine2 receptor antagonists separately. For studies that provided stratum-specific estimates, we combined them by means of the inverse-variance method. We included randomized controlled trials comparing acid-suppressive drugs (intervention) with either placebo or sucralfate control, as we were interested only in the influence of acid suppression on pneumonia.
Two investigators (C.S.E., J.W.L.) independently evaluated the eligibility of all studies retrieved from the databases on the basis of the predetermined selection criteria. They resolved any disagreements by discussion or in consultation with the co-corresponding authors (S.M.P., K.S.L.).
We assessed the methodologic quality of observational studies with the Newcastle–Ottawa Scale18 and that of randomized controlled trials with the Jadad scale19 (Appendix 1, available at www.cmaj.ca/cgi/content/full/cmaj.092129/DC1). We conducted subgroup analyses according to methodologic quality (low-quality studies v. high-quality studies). For the observational studies, low quality was defined as Newcastle–Ottawa Scale score ≤ 8.0 and high quality as score > 8.0 (maximum score 9). For the randomized controlled trials, low quality was defined as Jadad scale score ≤ 3.0 and high quality as score > 3.0 (maximum score 5).
We computed a pooled OR and 95% CI from the adjusted ORs and 95% CIs reported in the observational studies. For randomized controlled trials, we computed the summary relative risk from the relative risks of the individual trials using Mantel–Haenszel weighting.
We examined heterogeneity in results across the studies using Higgins I2 value, which measures the percentage of total variance in the summary estimate due to between-study heterogeneity.20
In light of the heterogeneity of study designs and population characteristics, we calculated the summary effect by means of the DerSimonian–Laird method21 for random-effects models.
We identified a total of 2377 articles in the initial search for observational studies, and we reviewed 60 abstracts and 18 full articles. We included 8 of these articles in our analysis. We identified 8513 randomized controlled trials, and we reviewed 914 abstracts and 35 full articles. We included 23 of these articles and 2 bibliographies of relevant articles in the study. In summary, we included five case–control studies,6,7,14,22,23 three cohort studies,2,10,24 and 23 randomized controlled trials25–47 in the final analysis (Figure 1).
Table 1 and Table 2 summarize the general characteristics of the 31 studies that were included in the analysis.2,6,7,10,14,22–47 The mean quality scores were 8.4 for the observational studies (maximum score 9) and 3.1 for the randomized controlled trials (maximum score 5).
The selected studies were published between 1985 and 2009. Five articles reported population-based studies,2,6,7,14,23 and 26 articles, including the 23 randomized controlled trials, reported hospital-based studies.10,22,24–47 Of the observational studies, five evaluated the association between use of acid-suppressive drugs and risk of community-acquired pneumonia,2,6,7,14,23 and three evaluated the association between use of these drugs and risk of hospital-acquired pneumonia.10,22,24
Meta-analyses for observational studies with the two types of acid-suppressive drug showed significant positive associations between use of proton pump inhibitors and risk of pneumonia (adjusted OR 1.27, 95% CI 1.11–1.46, I2 90.5%) and between use of histamine2 receptor antagonists and risk of pneumonia (adjusted OR 1.22, 95% CI 1.09–1.36, I2 0.0%) (Figure 2).
Meta-analysis of the randomized controlled trials examining risk of hospital-acquired pneumonia in association with use of histamine2 receptor antagonists confirmed the findings of the observational studies (relative risk 1.22, 95% CI 1.01–1.48, I2 30.6%) (Figure 3).
In subgroup analyses by type of pneumonia, we observed a significant positive association between use of proton pump inhibitors and community-acquired pneumonia (adjusted OR 1.34, 95% CI 1.14–1.57, I2 93.6%) and between use of histamine2 receptor antagonists and hospital-acquired pneumonia (adjusted OR 1.24, 95% CI 1.05–1.47, I2 0.0%) (Table 3).
Subgroup analyses by dose indicated a dose–response relationship. A higher dose of proton pump inhibitors was more strongly associated with pneumonia (adjusted OR 1.52, 95% CI 1.31–1.76, I2 27.5%) than the usual dose (adjusted OR 1.37, 95% CI 1.08–1.74, I2 86.5).
Subgroup analyses by duration of exposure showed that the strength of the association between use of proton pump inhibitors and risk of pneumonia decreased with longer duration of therapy before the index date (date of diagnosis of pneumonia). There were significant positive associations between risk of pneumonia and use of proton pump inhibitors within 7 days before the index date (adjusted OR 3.95, 95% CI 2.86–5.45, I2 0.0%), within 30 days before the index date (adjusted OR 1.61, 95% CI 1.46–1.78, I2 30.6%) and from 30 to 180 days before the index date (adjusted OR 1.36, 95% CI 1.05–1.78, I2 84.3%). The risk of pneumonia was greater with the use of histamine2 receptor antagonists within 7 days before the index date (adjusted OR 5.21, 95% CI 4.00–6.80, I2 not available). The risk also appeared greater with the use of these drugs within 30 days before the index date (adjusted OR 1.49, 95% CI 0.82–2.72, I2 80.4%) and from 30 to 180 days (adjusted OR 1.21, 95% CI 0.94–1.56, I2 27.6%), but these associations were not statistically significant.
Subgroup analyses of the 23 randomized controlled trials by comparators showed a significant positive association between use of histamine2 receptor antagonists and risk of pneumonia in studies that employed sucralfate as a control (relative risk 1.33, 95% CI 1.04–1.69, I2 24.7%). Placebo-controlled studies also indicated an overall increase in the risk of pneumonia with these drugs, but the result was not statistically significant (relative risk 1.09, 95% CI 0.80–1.48, I2 37.9%).
We conducted subgroup meta-analyses of the observational studies and randomized controlled trials by methodologic quality. Among the observational studies, we observed a significant positive association for both high-quality studies (adjusted OR 1.29, 95% CI 1.17–1.42, I2 0.0%) and low-quality studies (adjusted OR 1.15, 95% CI 1.00–1.32, I2 82.1%). Among the randomized controlled trials, the risk of pneumonia appeared greater in low-quality studies (relative risk 1.35, 95% CI 1.10–1.67, I2 12.5%), whereas there was no effect among the high-quality studies (relative risk 0.96, 95% CI 0.65–1.43, I2 47.0%).
Our results suggest that the use of acid-suppressive drugs is associated with an increased risk of pneumonia. Given the widespread use of acid-suppressive drugs, the implications of this increased risk are serious. If we assume that 19.7 cases of pneumonia occur for every 1000 individuals not receiving acid-suppressive drugs who are admitted to hospital,24 and if we also assume a 1.22- to 1.27-fold increase in the risk of pneumonia due to acid-suppressive drugs, as determined in this study, 24 or 25 cases of pneumonia can be expected for every 1000 recipients of these drugs. This translates to about one case of pneumonia for every 200 inpatients treated with acid-suppressive drugs. Given that 40%–70% of patients admitted to hospital receive acid-suppressive drugs,48 a considerable burden of morbidity and mortality of hospital-acquired pneumonia may be attributable to this type of therapy. In the context of community-acquired pneumonia, the impact of these drugs could be even more serious.
Several lines of evidence point to the biological plausibility of these observations. First, acid-suppressive drugs may increase the risk of pneumonia by inhibiting the secretion of gastric acid, thus allowing bacterial overgrowth and colonization in the upper alimentary tract and subsequent translocation to the lungs by aspiration.6,7,49 Second, hydrogen potassium adenosine triphosphatase is present not only in the parietal cells of the stomach, but also in the respiratory tract.50,51 It is conceivable that use of a proton pump inhibitor could alter the pH of the seromucinous secretions by inhibiting this enzyme, thereby encouraging bacterial growth in the respiratory tract, which could in turn lead to increased risk of pneumonia.52 Third, in vitro studies have shown that acid-suppressive drugs may impair the function of neutrophils and the activity of natural killer cells.53–59
Interestingly, the most striking increase in the risk of pneumonia in association with proton pump inhibitors was observed in the first week of use. The risk of pneumonia in association with use of proton pump inhibitors was attenuated, but still significant, between 30 and 180 days. Recipients of histamine2 receptor antagonists between 30 and 180 days before the index date appeared to have an increased risk of pneumonia, but the association was not statistically significant. These findings might reflect tolerance.52 Tolerance to histamine2 receptor antagonists generally develops within two weeks with repeated administration, resulting in a decline in acid suppression.60 Another reason may be that those who are more susceptible to pneumonia become ill with this disease early after starting acid-suppressive drugs, leaving fewer such individuals among those using these drugs for longer periods. That is, patients who remain on the drug are those who can tolerate it, whereas those who are susceptible select themselves out of the population at risk. This depletion of susceptible effect has been considered in other pharmacoepidemiologic studies of adverse events.61
Previous meta-analyses11,13,15 examined the effect of acid-suppressive drugs on pneumonia as a secondary outcome in randomized controlled trials. Cook and associates11 showed that the rate of pneumonia was higher among patients taking histamine2 receptor antagonists than among controls, but the difference was not statistically significant (OR 1.25, 95% CI 0.78–2.00). Conversely, Messori and colleagues13 found no difference in the risk of pneumonia between those who were given ranitidine and those who were given placebo (OR 0.98, 95% CI 0.56–1.72). However, they found an increased risk of nosocomial pneumonia in studies comparing ranitidine and sucralfate (OR 2.21, 95% CI 0.86–5.65). Finally, Pongprasobchai and coworkers15 reported that the incidence of nosocomial pneumonia did not differ between patients receiving proton pump inhibitors and those receiving histamine2 receptor antagonists. Compared with the previous meta-analyses, our review included more studies, which led to greater power to detect an effect. We also included observational studies, which enrolled a greater diversity of individuals, especially those taking high doses of acid-suppressive drugs.
Our analysis incorporated all relevant studies that we could identify to August 2009, including both observational and randomized controlled trials. We were also able to identify sources of heterogeneity by stratifying analyses on key variables.
Despite these strengths, our study had some limitations. First, we included only English-language publications for the selection of observational studies. We performed a subsequent search for all relevant observational studies without any language restrictions and found about 18% more citations. However, none of these articles met the inclusion criteria. It is unlikely that the language of the studies would have altered the validity or magnitude of the associations between acid-suppressive drugs and pneumonia. Second, the presence of gastroesophageal reflux disease might be a confounder,49 as those who receive acid-suppressive drugs often experience this condition, which in itself could be a risk factor for pneumonia. However, given that the included studies adjusted for factors such as comorbidities and other medications, any resulting bias was unlikely to have been great enough to explain the observed effect. Third, although the high-quality observational studies showed a significant effect, the high-quality double-blinded randomized controlled trials did not show a significant effect. This discrepancy might be attributable to methodologic rigour, but differences in study characteristics may also have contributed to the heterogeneous results.
Clinicians should carefully consider any decision to prescribe acid-suppressive drugs, especially for patients who are already at risk for pneumonia.62 Since it is unnecessary to achieve an achlorhydric state in order to resolve symptoms, we recommend using the optimal effective dose of the drug necessary to achieve desired therapeutic goals.
Competing interests: None declared.
This article has been peer reviewed.
Contributors: As the primary author, Chun-Sick Eom was responsible for the initial research plan; design of the study; collection, extraction and interpretation of the data; drafting of the manuscript; and statistical analysis. Christie Y. Jeon provided insight into the statistical methods and participated in drafting and critical revision of the manuscript for important intellectual content. Ju-Won Lim and Eun-Geol Cho were responsible for collection and extraction of the data and critical revision of the manuscript. Sang Min Park and Kang-Sook Lee were responsible for interpretation of the data, drafting of the manuscript and critical revision of the manuscript for important intellectual content, as well as contributing equally as co-corresponding authors. All authors read and approved the manuscript submitted for publication. This paper’s contents are solely the responsibility of the authors. Chun-Sick Eom is the guarantor for this paper and has full responsibility for this study.
Funding: This research was supported by the Basic Science Research Program of the National Research Foundation of Korea (grant 2010-0004429), which is funded by the Ministry of Education, Science, and Technology of the Korean government.