It is now widely recognized that the incidence of Clostridium difficile-associated disease (CDAD) and its severity have risen markedly since 1990. This is reflected in the rising diagnosis of both community-acquired and hospital-acquired cases and in their severity, as judged by frequent identification of several hypervirulent strains, higher rates of recurrence, resistance to antibiotics, complications (toxic megacolon, septic shock, intestinal perforation), prolongation of hospital stay, morbidity, case fatality, and overall mortality.1–4 In the face of declining use of antibiotics, these risks have increased 3- to 5-fold, particularly in the elderly, but there has also been a major rise in community-acquired disease5 involving younger and healthier cases: in some recent studies community-acquired infections have accounted for 40% or 50% of the total, with upward of 40% of patients having no history of prior exposure to antibiotics.5 A large number of risk factors have been identified, many of which are changing.3 In this issue of the journal, Kim et al6 provide startling information that recurrence of CDAD in patients taking a proton pump inhibitor (PPI) for at least 1 week after resolution of their bowel symptoms was 47.6%. This was compared with 4.8% in CDAD patients without recurrence, an odds ratio (OR) of 18.8 [95% confidence interval (CI): 2.05, 161.38; P=0.009]. The number of cases involved was small, but it adds this study to a rapidly growing number of reports identifying PPI use as a risk factor for CDAD. What does this mean?
With over 30 papers of varying quality already published on the association of PPI use with development or recurrence of CDAD, it behooves us to seek evidence for causality that may emerge from essentially epidemiologic studies. Space does not allow a detailed application of Bradford Hill’s classic criteria for causation,7 but a few of the more important of these are of particular relevance, namely, temporal relationship, strength of association, consistency, dose-response relationship, and biological plausibility.
Although PPIs first became available in the late 1980s, initial use was relatively small, and no consistent International Classification of Diseases coding existed for C. difficile infection (CDI) before 1993. Since that time PPI use has risen enormously and so has the incidence of CDI, but the rise of the latter was modest until about 2000, after which the rise was steep.1 Although studies have drawn attention to an apparently close temporal association between PPI use and CDAD,8,9 it must be noted that, around the same period that PPIs became generic, available over-the-counter and the third most used drug in the United States, fluroquinolone antibiotics, especially levoquin, were also being widely used, with C. difficile becoming increasingly resistant to them. More virulent toxigenic strains were also being increasingly identified, particularly in institutional outbreaks. Transmission of CDI by hospital personnel was not widely emphasized then. It is therefore difficult to exclude confounding of the temporal association by these or other factors. Nevertheless, in one carefully executed observational study of the incidence of CDAD among 76,908 hospitalized patients in New Jersey between 2001 and 2005 (inclusive), incidence increased from 5.08 to 8.42 cases/103 admissions (P<0.0005).9 During the time period, there was a decline in the use of H2-receptor antagonists (H2RAs), with a concomitant increase in the use of PPIs for prophylaxis of stress ulcers. In 2005 (the last year of the study), a carefully designed case-control analysis showed that PPI use, either before or during admission, was associated with CDAD, with an adjusted OR of 2.75 (95% CI: 1.68, 4.52): the association with H2RAs was not significant: adjusted OR.0.95 (95% CI: 0.39, 2.34). It is noteworthy that the usage of PPI correlated exactly with the increased incidence of CDAD (rs=1.0; P=0.017). Other studies that examine this effect, pooling antacids, H2RAs, and PPIs as “gastric acid-reducing drugs,” should not be used to examine the true risk of PPIs, as the effects of PPIs are multiple and not yet proven to be mediated solely or partly through reduction of acidity.
Relevant to the strength of association, a recent paper by Bavishi and DuPont10 provided an extensive systematic review of 30 studies before May 1, 2011, including 27 studies on PPI use and risk of developing CDC and 3 studies on risk of CDC recurrence. Another 5 papers have followed since then.6,11–14 From the review, 18 of 27 studies provided positive evidence of a positive association with adjusted relative risks (RR) ranging from 1.2 to 5.0, and the 3 retrospective cohort studies also found significant associations with recurrent CDI, with RRs between 1.4 and 4.2.10 In these studies there was also significant association with H2RA use in 5 of 13 but no association in the remaining 8. There was little or no information on doses, routes of administration, or durations of exposure to PPIs or H2RAs in most of these studies, variations in which probably accounted for some of the variation in RRs. The authors noted that studies failing to show an association involved predominantly hospitalized patients greater than 65 years of age, perhaps because many in this age group are already achlorhydric (eg, due to Helicobacter pylori gastritis) or may already have had high vulnerability to CDI (other major risk factors or comorbidity), and the effect of adding a PPI would have little additional effect on risk.
Among the remaining studies, a large prospective study of 4143 patients admitted to 6 hospitals in Canada identified 117 (2.8%) and 123 (3.0%) patients with health care-associated CDI and colonization, respectively: old age, antibiotic use, and PPI use were significantly associated with CDI. With PPI use, the OR was 2.64 (95% CI: 1.71, 4.09) for CD infection and 1.71 (95% CI: 1.15, 2.53) for CD colonization.11 A study of 10,154 hospitalizations and 241 cases of CDI from Buffalo, NY, found PPI use independently associated with an increased risk for CDI (adjusted hazard ratio=4.5 with 95% CI: 2.3, 9.0).12 Among hospitalizations during which 1, 2, 3, and 4 or more antibiotics were used, the adjusted hazard ratios for PPI use were 15.7, 4.9, 4.3, and 2.7, respectively, a finding that awaits explanation. A study from North Carolina13 claimed “No role for proton pump inhibitors,” but the claim is misleading and hard to interpret: it combined data for PPIs and H2RAs—which have been shown as having no effect in 8 of 13 studies10—and furnished no information on drug administration for PPIs versus H2RAs in included patients. A significant effect of PPIs could have been masked by lack of effects in patients on H2RAs.13 Final findings come from a large, population-based, retrospective, nested, case-control study of community-based infection in Iowa.14 Over 4 years, 684 cases of CDI were identified: 304 were community acquired (CA-CDI), 338 were hospital acquired, and 42 were indeterminate. Over 18% of CA-CDI and 5% of controls (10 per case) received “gastric acid suppressants” in the 180 days before diagnosis, an adjusted OR of 2.3 (95% CI: 1.56, 3.39). The study is again somewhat flawed by combining PPIs and the much less hazardous H2RAs, so that the true OR for PPI alone may be considerably higher. In this study, 84% of those affected also received one or more antimicrobial agents in addition to an acid-reducing drug. The interactions of different PPIs and multiple different antibiotics are potentially complex and not discernable from these studies.
Putting all these together it appears that 21 of 30 studies (70%) on the development of CDAD and 4 of 4 studies on disease recurrence (100%) show that PPI use is a real risk factor. The association is strong but of uncertain magnitude, the risk probably varying with differences in other risk factors, case mix, and study design. The Public Health Agency of Canada website contains a warning that PPIs may increase the risk of CDI. The association is broadly consistent, although imperfect. No clear association of H2RAs emerges but remains possible, with variations in dose of drug and route of administration possibly accounting for the heterogeneity in the results of published studies.
This brings us to the dose-response relationship. Assuming for the moment that elevation in gastric pH is responsible for at least some of the effect of PPIs, gastric juice is strongly bactericidal at pH<4.0, less effective at pH 5.0, and ineffective at pH 6.0.10 The fact that PPIs, which in commonly used doses frequently elevate pH into this range, emerge as a risk factor for CDI, whereas H2RA use that rarely leads to similar pHs is much less hazardous, argues indirectly that gastric pH is a relevant variable, but this does not deal directly with dose response to PPI. There also appears to be a requirement for physiologic concentrations of bile salts for C. difficile spores to germinate in gastric juice.10,15,16 As PPIs, particularly in higher doses, delay gastric emptying, whereas 2 of the commonly used H2RAs—ranitidine and nizatidine—are prokinetic and may accelerate gastric and colonic emptying, the duration of spore delay in conditions of elevated pH or bowel transit may be another difference between PPIs and H2RAs.17–20 The best epidemiologic evidence for an acid suppression-mediated dose-response effect of PPIs on the risk of developing CDAD comes from a large pharmaco-epidemiologic study involving 101,796 hospital discharges from a tertiary care center over a 5-year period. In this study, the incidence of CDAD rose from 0.3% (95% CI: 0.21, 0.31) in patients receiving no acid-reducing drugs, to 0.6% (95% CI: 0.49, 0.79) in those receiving H2RAs, to 0.9% (95% CI: 0.8, 0.98) in those receiving PPI once daily, and to 1.4% (95% CI: 1.15, 1.72) in anyone receiving more frequent or higher doses.21 The large sample size and narrow confidence intervals for each risk estimate offer strong support to a dose-response effect. However, the incidence rates are not high, probably reflecting the heterogeneity of cases included just based on a discharge diagnosis.
The final issue to be addressed is biological plausibility. At first sight it may seem that the association is plausible, when attention is confined to the hypothesis that hypochlorhydria increases host susceptibility to bacterial infections, as many other enteric bacterial infections appear to have been increased by hypochlorhydria.22,23 However, there may be alternative or additional factors at work with PPI therapy. The target of PPI drugs is the enzyme H+/K+ ATPase, which, in addition to being present in the parietal cell, is also found in polymorphonuclear leukocytes, colonic epithelial cells, liver, bone, eyes, and other sites, where many of its roles are still unknown. In addition, as reviewed elsewhere,10 numerous experimentally demonstrated effects of PPIs (anti-infective, anti-inflammatory, and immuno-modulatory in nature) could all theoretically affect host susceptibility to CDI. These include antioxidant effects (especially in processes involving HOCl−, iron, or copper), increased hydroxyl scavenging, inhibition of neutrophil functions (calcium transport, microbial phagocytosis, bactericidal activity, expression of adhesion molecules), and chemotactic migration: none of these effects have been well studied in humans. However, at the core of the current hypothesis is the surmise that gastric hypoacidity alters host defense against C. difficile. Hypochlorhydria is an established risk factor for other enteric infections, including Escherichia coli, Campylobacter jejuni, and Vibrio cholera10,22,23 so that postulating that this may occur with C. difficile may seem just a simple extension of an accepted hypothesis.
However, it is far from clear how the majority of CDI is acquired or how CDAD develops. The organism C. difficile exists in 2 forms, an acid-sensitive vegetative form and an acid-resistant spore form, believed to be the main vector of the disease. Both forms are excreted in stool (vegetative forms being 10-fold more common than spores) and transmitted by fecal contamination of hands or fomites, by persons who have touched the skin or fomites of infected subjects, and by carriage between patients/beds by medical personnel. Medical personnel can be exposed during outpatient contacts with unsuspected cases. Most recently, aerosolization of the bacterium after flushing toilets has been shown to contaminate surrounding surfaces and air, with currently unknown consequences: closing the toilet lid before flushing avoids the hazard.24
Although spores survive gastric acidity, it is known that the vegetative forms are normally destroyed in the stomach; however, vegetative cells remain a potential source of infection, especially when gastric pH is >5.0. Furthermore, for spores to germinate, pH must also be high. Specific bile salts and glycine act as cogerminants of C. difficile spores, converting them to vegetative (infectious) forms: glycine probably alters the ratio of deconjugated to conjugated bile salts in the lumen.25 The effects of bile salts are complicated.26 Secondary bile salts, for example, deoxycholate, prevent vegetative growth. In health, spores survive gastric acid, reaching the small intestine where they can germinate in high concentrations of cholate derivatives and glycine, but normal concentrations of the secondary bile salt deoxycholate inhibit vegetative growth. After severe disruption of gastrointestinal microbial populations by PPI-induced hypochlorhydria,27,28 secondary bile salts such as deoxycholate are much reduced, allowing unrestricted vegetative growth. Chenodeoxycholate, a primary bile salt, inhibits germination and protects the small intestine by inhibiting production of vegetative forms, until spores reach the anerobic environment of the large bowel. Physiologically, chenodeoxycholate is rapidly absorbed in the ileum, generating conditions favorable to germination. These conditions are reversed by cholestyramine, which binds bile salts that stimulate spore germination.26 Bile salts are found in gastric juice in most GERD patients, the major users of PPIs. The normally protective anaerobic milieu of the colon is greatly disturbed by antibiotics such as fluoroquinolones and clindamycin,26 and these conditions, added to the floral disturbances caused by PPIs, increase the risk of CDC.
In contrast, the colon is the site of CDAD. Two points seem relevant. The target enzyme of any PPI is present in the colon29: inhibiting it may increase cellular susceptibility to infection. Furthermore, PPIs have been shown to disrupt gut ecology. Bacterial counts of many species increase in gastric hypochlorhydria, with PPIs clearly affecting both small intestinal and colonic microflora27,28: normal flora is critically involved in mucosal defenses against inflammation. Beyond these points, PPIs affect gastric acidity, motility, mucus viscosity, tight-junction permeability, and bacterial translocation, and any or all of these processes could be involved in causing CDAD, the pathogenesis remaining far from clear. Thus, there are multiple factors that plausibly link PPI activity with increasing risks of CDAD, but the principal biological mechanism on which this plausibility rests remains to be clearly identified. This is beyond the reach of epidemiologic studies.
It thus appears from a large number of studies that there is epidemiologic evidence, substantial by accepted criteria, that PPI use is a risk factor for CDI, CD colonization, and especially CDAD. The magnitude of risk in the individual case remains uncertain: from the more recent, large, carefully executed studies, the hazard ratio is generally between 2.5 and 5.0 but may be much higher in some circumstances. There is a suggestion that the magnitude of this risk may be higher for recurrence of the disease than for initial acquisition, but this could be different in community-acquired and hospital-acquired diseases. Whenever possible, and as soon as possible, use of PPIs should be stopped in CD-infected patients, especially in those in whom no clear indication for treatment exists, in all those leaving intensive care units, and particularly in those leaving the hospital. Although it is easy to recommend large prospective further studies, to design them in a way that avoids confounding remains a challenge. Even with powerful techniques, such as multivariate logistic regression, variations in comorbidity, numbers and types of antibiotics used, PPI dosage, and route of administration are likely to yield results that defy precise interpretation. Similar issues surround H2RA use, although the associated risk, if any, appears less. The very high hazard ratios occasionally observed, as in the present study,6 are unlikely to apply to studies broadly inclusive of cases but may apply in some subgroups as yet undefined. The observed increased risk provides one more reason that PPI use should be confined to diagnostically clear indications, used in the minimum effective dose for the shortest possible time, and promptly stopped in those with C. difficile colitis.
1. Elixhauser A, Jhung M Clostridium difficile-Associated Disease in US Hospitals, 1993-2005: statistical brief #50 Health Care Cost and Utilization Project (HCUP): statistical brief #50 (internet). 2006–2008 PMID 21735570 Agency for Healthcare Policy and Research (US)
2. Viswanathan VK, Mallozzi MJ, Vedantam G. Clostridium difficile
infection: an overview of the disease and its pathogenesis, epidemiology and interventions. Gut Microbes. 2010;1:234–242
3. Lo Vecchio A, Zacur GM. Clostridium difficile
infection: an update on epidemiology, risk factors, and therapeutic options. Curr Opin Gastroenterol. 2012;28:1–9
4. Carroll KC, Bartlett JG. Biology of Clostridium difficile
: implications for epidemiology and diagnosis. Ann Rev Microbiol. 2011;65:501–521
5. Khanna S, Pardi DS, Aronson SL, et al. The epidemiology of community-acquired Clostridium difficile
infection: a population-based study. Am J Gastroenterol. 2012;107:89–95
6. Kim YG, Graham DY, Byung IJ. Proton pump inhibitor use and recurrent Clostridium difficile-
associated disease: a case-control analysis matched by propensity score. J Clin Gastroenterol. 2012;46:397–400
7. Bradford-Hill A. The environment and disease: association and causation. Proc R Soc Med. 1965;58:295–300
8. Dial S, Delaney JAC, Barkhun AN, Suissa S. Use of gastric acid-suppressive agents and the risk of community acquired Clostridium difficile
-associated disease. JAMA. 2005;294:2989–2995
9. Jayatilaka S, Shakov R, Eddi R, Bakaj G, et al. Clostridium difficile
infection in an urban medical center: five-year analysis of infection rates among adult admissions and association with the use of proton pump inhibitors. Ann Clin Lab Sci. 2007;37:241–247
10. Bavishi C, DuPont H. Systematic Review: the use of proton pump inhibitors and increased susceptibility to enteric infection. Aliment Pharmacol Ther. 2011;34:1269–1281
11. Loo VG, Bourgault A-M, Poirier L, et al. Host and pathogen factors for Clostridium difficile
infection and Colonization. N Engl J Med. 2011;365:1693–1703
12. Stevens V, Dumyati G, Brown J, et al. Differential riskof Clostridium difficile infection with proton pump inhibitor use by level of anti-biotic exposure. Pharmacoepidemiol Drug Saf. 2011;10:1035–1042
13. Naggie S, Miller BA, Zuzak KB, et al. A case-control study of community-associated Clostridium difficile infection: no role for proton pump inhibitors. Am J Med. 2011;124:276e1–276e7
14. Kuntz JL, Chrischilles EA, Pendergast JF, et al. Incidence of and risk factors for community-associated Clostridium difficile
infection: a nested case-control study. BMC Infect Dis. 2011;11:194–200
15. Wheeldon LJ, Worthington T, Hilton ACV, et al. Physical and chemical factors influencing the germination of C.difficile
spores. J Appl Microbiol. 2008;105:2223–2230
16. Nerandzic MM, Pultz MJ, Donskey CJ. Examination of potential mechanisms to explain the association between proton pump inhibitors and Clostridium difficile
infection. Antimicrob Agents Chemother. 2009;53:4133–4137
17. Sanaka M, Yamamoto T, Kuyama Y. Effects of proton pump inhibitors on gastric emptying: a systematic review. Dig Dis Sci. 2010;55:2431–2440
18. Laine-Cessac P, Turcant A, Premel-Cabic A, et al. Inhibition of cholinesterases by histamine 2 receptor antagonist drugs. Res Comm Chem Pathol. 1993;79:185–193
19. Sun WM, Hasler WL, Lien HC. Nizatidine enhances the gastro-colonic response and the colonic peristaltic reflex in humans. J Pharmacol Exper Ther. 2001;299:159–163
20. McCallum RW, Zarling EJ, Goetsch AC, et al. Multicenter, double blind , placebo- controlled, crossover study to assess the acute prokinetic effect of nizatidine controlled-release (150 and 300 mg) in patients with gastroesophageal reflux disease. Am J Med Sci. 2010;340:259–263
21. Howell MD, Novack V, Grgurich P, et al. Iatrogenic gastric acid suppression and the risk of nosocomial Clostridium difficile
infection. Arch Intern Med. 2010;170:784–790
22. Dial MS. Proton pump inhibitors and enteric infections. Am J Gastroenterol. 2009;104(S2):S10–S16
23. McCarthy DM. Adverse effects of proton pump inhibitor drugs: clues and conclusions. Curr Opin Gastroenterol. 2010;26:624–631
24. Best EL, Sandoe JAE, Wilcox MH. Potential for aerosolization of Clostridium difficile after flushing toilets: the role of toilet lids in reducing environmental contamination risk. J Hospital Infect. 2012;80:1–5
25. Thielsen J, Nehra D, Citron D, et al. Suppression of gastric acid secretion in patients with gastroesophageal reflux disease results in gastric bacterial overgrowth and deconjugation of bile acids. J Gastrointest Surg. 2000;4:50–54
26. Burns DA, Heap JT, Minton NP. Clostridium difficile
spore germination: an update. Res Microbiol. 2010;161:730–734
27. Kanno T, Matsuki T, Oka M, et al. Gastric acid reduction leads to an alteration in lower intestinal microflora. Biochem Biophys Res Com. 2009;381:666–670
28. Wallace JL, Syer S, Denou E, et al. Proton pump inhibitors exacerbate NSAID- induced small intestinal injury by inducing dysbiosis. Gastroenterol. 2011;141:1314–1322
29. Kaunitz JD, Sachs G. Identification of a vanadate-sensitive potassium-dependent proton pump from rabbit colon. J Biol Chem. 1986;261:14005–14010