Acid Suppression and Pneumonia: Title and subTitle BreakA Clinical Indication for Rational Prescribing

Disorders of the upper gastrointestinal (GI) tract, both symptomatic and asymptomatic, have been ubiquitous in humans throughout the history of medicine. Although important advances, such as the discovery of Helicobacter pylori, have offered curative options for many patients, lifestyle habits and the increasing use of other medication classes, including nonsteroidal anti-inflammatory drugs (NSAIDs) and anticoagulants in western countries, ensure that disorders such as gastroesophageal reflux and NSAID gastropathy will remain prevalent for decades to come.

The development over the last 40 years of potent acid-suppressing agents, first H₂ receptor antagonists (H₂RAs) and subsequently the more potent proton pump inhibitors (PPIs), have led to major improvements in both disease management and diagnosis. Because of their efficacy and impressive safety profile that has far exceeded original expectations, acid-suppressing drugs have consistently been among the most widely prescribed medications worldwide, with almost $13 billion in sales in 1998 and an annual growth rate of 3%.¹ From the relief of ulcerative esophagitis to noncardiac chest pain, the consistent benefit realized from this class in general and PPIs in particular have led to their widespread adoption as the therapy of first choice² and, in many cases, the initial diagnostic test considered.³

Concerns about long-term use of these agents, including masking of preexisting malignancy, increasing susceptibility to enteric infection, malabsorption of nutrients, and development of gastrointestinal neoplasia, have, for the most part, failed to be realized.⁴ With this reassurance, and a deluge of industry-sponsored studies showing impressive efficacy in both the treatment and the prevention of disease, it is not surprising that both prescription and over-the-counter versions of even the more potent agents are now widely advertised to the public.

In this issue of JAMA, Laheij and colleagues⁵ use a well-defined database containing the medical records of half a million Dutch patients to investigate whether chronic acid inhibition is associated with the development of community-acquired pneumonia, presumably mediated through increased colonization of the stomach with potentially pathogenic bacteria. Although the database is large and the information is detailed and free of intentional bias, the authors acknowledge that the potential confounding by indication was far too great to allow the primary outcome to rest on the overall case-control study alone. Therefore the authors also conducted a separate “nested” analysis of patients with pneumonia who had been administered acid-suppressive drugs at the time of contracting pneumonia and in whom exposure was only prior to developing the disease.

After controlling for age and sex in the nested case-control analysis, the investigators found that patients taking acid-suppressing drugs were almost twice as likely to develop pneumonia as were those who had previously discontinued the medication. They calculated that this translated into approximately 1 case of pneumonia for every 100 years of patient exposure. Of interest is that this risk is roughly comparable in magnitude to the risk of upper GI bleeding attributed to NSAIDs, a problem that has been the subject of considerable clinical research.⁶

The results of this study are noteworthy and satisfy many of the accepted criteria for causation. The association is reasonably strong and the confidence intervals do not approach unity. The biological plausibility of acid inhibition leading to bacterial overgrowth is certainly not new. Patients with gastroesophageal reflux taking PPI therapy appear to have a significantly increased load of intragastric bacteria compared with those in whom such therapy has been discontinued, particularly if gastric pH is maintained at optimally therapeutic levels (≥4).⁷ It has been difficult to demonstrate the clinical importance of this finding but by no means has such significance been excluded. Although a large well-designed study comparing sulcrafate with ranitidine for stress ulcer prophylaxis in critically ill patients did not demonstrate a statistically significant increase in ventilator-associated pneumonia, a trend toward increased pneumonia with ranitidine was observed.⁸

Laheij and colleagues also went to considerable effort to ensure that the exposure, in this case acid-suppressing drugs, did indeed precede the outcome of pneumonia and to properly classify patients as to whether the exposure was present at the time of pneumonia diagnosis. They also examined separately the effects of the less-potent H₂RAs and different PPIs. In doing so, the authors were able to demonstrate a tantalizing dose-response effect, as the relative risk of pneumonia was modest with H₂RAs, greater with PPIs, and greater still for those taking more than 1 defined daily dose of a PPI.

It is unlikely that a randomized controlled trial evaluating PPI use with pneumonia as the primary outcome measure will ever be conducted. A large randomized trial that assessed upper GI complications of NSAIDs required approximately 9000 patients to demonstrate that a 1% difference in complications was statistically significant.⁹ Similar numbers would likely be needed to confirm or disprove the association between acid-suppressive agents and pneumonia noted in this study. Given this estimate and the paucity of suitable alternatives for patients with conditions such as chronic gastroesophageal reflux, case-control or retrospective cohort studies will likely be the best evidence available.

The main criticism of the study by Laheij et al is similar to that of all retrospective studies. Even extensive risk factor matching and statistical adjustment cannot control for unknown important differences between groups. In addition there is always a nagging concern that patients who acquired pneumonia had more severe reflux, as suggested perhaps by their inability to discontinue acid-suppressing medications, and therefore would be at higher risk for aspiration and ultimately pneumonia regardless of medication use.

Despite these concerns, the results of this study are worth noting, particularly as the indications for chronic acid suppression continue to expand. These drugs, particularly PPIs, are now recommended for the treatment of diseases such as reflux esophagitis and are frequently advocated for the prevention of such problems as NSAID gastropathy¹⁰ and even esophageal dysplasia in Barrett esophagus.¹¹ As the indications for these drugs expand further and as the population ages, the number of patient-years of exposure will continue to increase and any unrecognized complications will become of greater importance.

If acid suppression causes some cases of pneumonia, it is reassuring that the risk is relatively small and that the complication in most cases is usually amenable to therapy. However, no medication is without potential adverse effects. Concerns for patient safety should guide initial prescribing and perhaps more importantly, chronic use of even the most apparently benign drug.