PPIs = CAP? The Possible Connection
By David Kiefer, MD. Dr. Kiefer is Clinical Instructor, Family Medicine, University of Washington, Seattle; Clinical Assistant Professor of Medicine, University of Arizona, Tucson; Adjunct Faculty, Bastyr University , Seattle; he reports no financial relationship to this field of study.
Synopsis: Protein pump inhibitors cause changes in gastric acid that promote bacterial colonization of the upper gastrointestinal tract and may be connected with an extra risk of community-acquired pneumonia, though the research reviewing this effect in humans has been mixed.
Proton pump inhibitors (ppis) are one of the important tools in the treatment of several upper gastrointestinal conditions, including gastroesophageal reflux disease, gastritis, and peptic ulcer disease. The clinical use of this class of medication varies, from short-term (one to several months) to more chronic use, but they are widely prescribed; PPIs are among the most-prescribed medicine worldwide, accounting for $26.5 billion globally in 2008,1 and estimates are that approximately 40% of the United States population uses PPIs or H2-blockers (H2Bs).2 Recently there has been a slew of articles documenting the problems with the continued daily use of PPIs, especially concerning upper respiratory infection (URIs) and community-acquired pneumonia (CAP). This article will review the mechanism of action of PPIs and the proposed association between chronic use and respiratory infections.
Both H2Bs and PPIs can increase gastric pH above 4, though PPIs do so more effectively.1,3,4,5 This pH threshold is necessary to help decrease upper gastrointestinal symptoms, but is also a level above which many bacteria are able to survive when otherwise they would have been killed by the acidity.1,3 It is possible that this less-acidic environment allows bacterial colonization of the stomach, and in some cases with bacteria that can colonize the respiratory tract and then cause pneumonia and other respiratory infections.5,6 For example, people taking PPIs have increased intragastric concentrations of aerobic bacteria, which may also be associated with micro-aspiration and pulmonary colonization.7 However, sometimes the bacteria found in the stomach of people taking PPIs differ from those collected during bronchoalveolar lavage, leading researchers to postulate that some mechanism other than aspiration may be involved.7
In addition, PPIs may cause impaired leukocyte function, including decreased adhesion to endothelial cells and reduced generation of reactive oxygen species involved in bacterial killing.5
Retrospective Case-control Research
One of the most common approaches to studying the PPI-CAP connection is via a case-control design. Essentially, cases of CAP in a given demographic are matched with a control group, and then statistical techniques are used to search for variables that most account for why CAP developed in the case group but not in the control group. Allowing the identification of important associations, these research approaches do not, however, pin down a cause-effect relationship; the latter is best left to randomized controlled trials (RCTs). With respect to confounding variables, there are several issues to consider when interpreting the results of the research discussed below. First, some research has found that the recent prescription of PPIs is more highly associated with CAP than distant or long-standing use; perhaps these medicines actually are being used for early, non-specific symptoms of respiratory disease, essentially a misclassification in statistical terms. Also, patients with multiple comorbid conditions may be more likely to be taking PPIs, be older, and be generally more likely to have respiratory infections,8 so teasing out the effect of PPIs can be difficult.
There were two initial attempts to explore this issue; these studies then formed the basis for a series of follow-up research. In one, researchers examined a large Dutch medical records database for an association between use of H2Bs or PPIs and CAP.3 A total of 364,683 people who were currently taking or who had recently taken and then stopped PPIs or H2Bs were involved over 2.7 years, and 5,551 cases of pneumonia were identified. Of this group, people currently using either H2Bs or PPIs were more likely to develop pneumonia than people who had stopped using acid-suppressing medication, with adjusted odds ratios (OR) of 1.63 (confidence interval [CI] = 1.07-2.48) and 1.89 (CI = 1.36-2.62), respectively. The researchers noted that the effect of PPIs, but not H2Bs, showed a dose-response effect, making a more convincing case for a true biological effect. As a purely observational research trial, uncontrolled confounders, and misclassification of exposure (prescription of medication was assumed to indicate compliance) and diagnosis (not all pneumonias were confirmed by radiographs) could have compromised results. Another unique, but possibly negative, feature of this trial was that there was no control group not taking any acid-suppressing medication.
Then, a Danish research group examined the demographics and confounding medical illnesses for people discharged from the hospital with a diagnosis of CAP, and matched this group with controls in the general population.9 The OR of CAP was 1.5 (CI = 1.3-1.7) for people taking PPIs; no association was found for H2Bs (OR = 1.1, CI = 0.8-1.3) nor for people who had taken PPIs in the past. This study was slightly different than the Dutch trial in that it only examined recently hospitalized cases of CAP.
A study in the United Kingdom (UK) was designed to follow up on the previously described studies and address criticisms of confounding variables, and to explain why the greatest risk may have been in recent dosing of acid-suppressing medications vs. people who had been taking the medication for longer periods of time. These researchers used a case-control study in a UK database, selecting 80,066 cases (first diagnosis of CAP during the study period) and 799,881 control patients (from the same general practice site) from a cohort of 7,347,764.10 Exposure was defined as any PPI use before or after the date of CAP diagnosis. After first analysis, the risk of CAP with PPI use was 2.05 (CI = 1.96-2.15, P < 0.001), though this association disappeared after adjusting for all possible covariates, such as comorbid conditions and any of a number of factors that could be associated with an increased risk of pneumonia (i.e., smoking). However, when the results were adjusted for people who had been taking PPIs for less than 30 days (and especially within two days), there was an increased risk of CAP of 1.74 and it remained significant even with covariates factored in (CI = 1.49-2.03; P < 0.001). The researchers make a convincing case that their efforts, which included analysis of potentially significant confounding variables, provided more accurate results than the preceding Dutch and Danish trials that both showed increased CAP risk with PPIs. The researchers also use the fact that there was a more pronounced PPI effect on CAP within a few days of starting the prescription, also seen in the Dutch and Danish trials, as evidence against a causal relationship; it seems more plausible that a "true" biological effect would occur in people who had longer-standing acid suppression and therefore decreased defenses against respiratory bacteria. A secondary analysis looked at risk of CAP with H2Bs and found greater increases in risk, a confusing result given that H2Bs are weaker inhibitors of acid production than are PPIs.
These initial attempts to clarify an association between PPIs and CAP have now been repeated by other research groups focusing on specific demographics. For example, in elderly patients who had survived hospitalization for pneumonia in the past, PPIs increased the risk for recurrent CAP by 12% (OR = 1.5, CI = 1.1-2.1) vs. 8% for non-PPI users, with all of the increased risk accounted for by people who had started PPIs after discharge as compared with those who had been taking PPIs before the original CAP event.11 Another research group examining 194 cases of CAP in elderly patients in a community hospital failed to find an association with current PPI use, though this study only had categories of use and non-use, whereas most other studies add a third "recent use" category, possibly affecting results.8 A case-control review of CAP records from 300 general practice clinics in the UK found an adjusted OR of 1.55 (CI = 1.36-1.77) for PPI use within 30 days; there was not an increase in CAP cases with H2B use.12 As with some of the other trials, there is the possibility of diagnosis misclassification, that the medications weren't actually taken (use was based on prescription), and limitations in diagnostic accuracy (the lack of radiograph confirmation of CAP).
Another trial found an increased risk of recurrent pneumonia in PPI users (RR = 1.16, CI = 1.03-1.31) when compared to nonusers when records from general practices in the UK were examined13; this risk only occurred if the PPIs were started within 12 months of the CAP event and there was a small dose-response effect. No increased risk was found for users of H2Bs. The authors postulated that the slightly increased risk demonstrated in this analysis might be due to underlying CAP associated with the GI disorders for which people were being treated with PPIs. Finally, in adults aged 65-94 a CAP case (n = 1,125) control (n = 2,235) study found no increased risk with PPI or H2B use (OR = 1.03, CI = 0.86-1.24), leading the authors to suggest that past positive findings in other trials might be entirely due to confounding variables.14 These researchers made a significant attempt to follow up on the pneumonia diagnosis to address the misclassification that has plagued other research; it is possible that this enhanced the accuracy of these particular results.
There are no PPI clinical trials with respiratory infections as the primary endpoint, so information about this presumed connection has to be gleaned from other types of trials. One meta-analysis of RCTs of PPIs primarily for GERD or PUD provides some initial data regarding the clinical association between PPIs and CAP beyond what we can learn from simple observational studies.6 The research group located 70 RCTs, only seven of which fit the strict inclusion criteria, mainly because most trials failed to specifically list patients with respiratory infections as adverse effects. Statistical analyses on the seven trials showed a trend toward an increased rate of respiratory infections with PPI use (13%), with five of seven trials demonstrating more infections in the PPI groups than the placebo groups; however, the overall trend was not statistically significant (OR = 1.42, CI = 0.86-2.35, P = 0.17). Of note, one clinical trial in this meta-analysis showed a statistically significant increase in respiratory infections when esomeprazole 40 milligrams daily was administered compared to placebo (P = 0.05). The authors of this meta-analysis calculated the number-needed-to-harm as 77; thus, a risk of minimal clinical relevance for the general population.
Another retrospective analysis of the adverse respiratory effects in 31 RCTs (16,583 patients total) of esomeprazole at 20 or 40 milligrams daily vs. placebo or other PPIs (lanzopresole or omeprazole) found no significant difference in the rate of respiratory infections, including pneumonia, in patients with gastric acid-related disorders.15 With this type of study, there is a possible selection bias as a result of only including patients with specific medical conditions, in this case, gastric acid-related disorders.
There is a significant body of information suggesting an association between H2Bs and pneumonia in the critical care setting,6 although some trials have either shown no connection7 or simply a non-statistically significant trend. One example is a cohort of 63,878 patients admitted to the hospital, excluding the intensive care unit, where PPI, but not H2B, use was associated with an increase in hospital acquired pneumonia (OR = 1.3, CI = 1.1-1.4).16 Sedatives and neuromuscular blockers, but not PPIs, were associated with an increased risk of pneumonia in 787 patients in the medical intensive care unit.17
PPI are a widely prescribed class of medications for a variety of gastrointestinal disorders. There are convincing mechanistic data for the effects of PPIs, via increased gastric pH and immune system changes, on bacterial colonization of the upper gastrointestinal tract and respiratory tract, which could translate into an increased risk of respiratory infections, including CAP. Most of the clinical research on PPIs and CAP risk comes in the form of either case-control research or retrospective analyses of adverse effects that occurred with the use of PPIs in RCTs on gastrointestinal disorders. In the former, results have been mixed, with some analyses showing a connection between PPI use (current, more than past) and CAP, with others showing no effect. In some cases, a dose-response effect was documented, and usually, but not always, the PPI effect was greater than that for H2Bs. Differing results could be due to variations in demographics (country, age, comorbidities, etc.) or research methodologies (the type of controls, identification of confounding variables, documentation of medication compliance or radiographically-confirmed CAP). Interestingly, some studies have shown an increased CAP risk with recently prescribed PPIs, a bit counterintuitive; different researchers, as explained above, have offered explanations for why this might have occurred or what it might mean. The second type of research methodology, a secondary retrospective analysis of adverse effects in trials using PPIs for gastrointestinal disorders, has failed to find a PPI-CAP connection. Most researchers addressing this topic, even when study results were negative, mention the importance of factoring in a possible respiratory connection with PPI use; the morbidity and mortality associated with CAP is not negligible, making it extremely important to be prescribing PPIs for only established indications.
Until the medical research completely resolves any possible PPI-CAP connection, it seems prudent to judiciously apply a risk-benefit analysis to a given patient's situation when making a decision to start or continue PPIs. The medical literature hints at a slightly higher CAP risk for people who have recently started PPIs, for elderly patients who previously had CAP and then started PPIs, and generally with PPIs employed at the higher end of their dose range. Extra care is, of course, warranted in considering these medications in people for whom CAP would be particularly devastating, such as the elderly. One researcher said it well: "...there is good evidence that (PPIs) are frequently being prescribed for non-specific and inappropriate reasons, and that a large number of patients are taking these agents for much longer than necessary...stop these agents in patients with previous dyspepsia but no endoscopic disease or gastric irritant use...(r)educe dose gradually over a period of a few weeks to avoid rebound acid secretion..."1 With this approach, clinicians would avoid some of the possible CAP risk with the use of PPIs.
1. Logan IC, Sumukadas D, Witham MD. Gastric acid suppressantstoo much of a good thing? Age Ageing 2010;39:410-411.
2. Restrepo MI, Mortensen EM, Anzueto A. Common medications that increase the risk for developing community-acquired pneumonia. Curr Opin Infect Dis 2010;23: 145-151.
3. Laheij RJ, Sturkenboom MC, Hassing RJ, et al. Risk of community-acquired pneumonia and use of gastric acid- suppressive drugs. JAMA 2004;292:1955-1960.
4. Vakil N. Acid inhibition and infections outside the gastrointestinal tract. Am J Gastroenterol 2009;104(Suppl 2):S17-20.
5. Wise MP, Saayman AG, Frost PJ. Acid-suppressive medication and hospital-acquired pneumonia. JAMA 2009;302:1416.
6. Sultan N, Nazareno J, Gregor J. Association between proton pump inhibitors and respiratory infections: A systematic review and meta-analysis of clinical trials. Can J Gastroenterol 2008;22:761-766.
7. Thomson AB, Sauve MD, Kassam N, Kamitakahara H. Safety of the long-term use of proton pump inhibitors. World J Gastroenterol 2010;16:2323-2330.
8. Gau JT, Acharya U, Khan S, et al. Pharmacotherapy and the risk for community-acquired pneumonia. BMC Geriatr 2010;10:45.
9. Gulmez SE, Holm A, Frederiksen H, et al. Use of proton pump inhibitors and the risk of community-acquired pneumonia. A population-based case-control study. Arch Intern Med 2007;167:950-955.
10. Sarkar M, Hennessy S, Yang YX. Proton-pump inhibitor use and the risk for community-acquired pneumonia. Ann Intern Med 2008;149:391-398.
11. Eurich DT, Sadowski CA, Simpson SH, et al. Recurrent community-acquired pneumonia in patients starting acid-suppressing drugs. Am J Med 2010;123:47-53.
12. Myles PR, Hubbard RB, McKeever TM, 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;18:269-275.
13. Rodríguez LA, Ruigómez A, Wallander MA, Johansson S. Acid-suppressive drugs and community-acquired pneumonia. Epidemiology 2009;20:800-806.
14. Dublin S, Walker RL, Jackson ML, 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;19:792-802.
15. Estborn L, Joelson S. Occurrence of community-acquired respiratory tract infection in patients receiving esomeprazole: Retrospective analysis of adverse events in 31 clinical trials. Drug Saf 2008;31:627-636.
16. Herzig SJ, Howell MD, Ngo LH, Marcantonio ER. Acid-suppressive medication use and the risk for hospital-acquired pneumonia. JAMA 2009;301:2120-2128.
17. Beaulieu M, Williamson D, Sirois C, Lachaine J. Do proton-pump inhibitors increase the risk for nosocomial pneumonia in a medical intensive care unit? J Crit Care 2008;23:513-518.