Association Between Proton Pump Inhibitor Therapy and Clostridium difficile Infection in a Meta-Analysis

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2012;10:225 233 Association Between Proton Pump Inhibitor Therapy and Clostridium difficile Infection in a Meta-Analysis ABHISHEK DESHPANDE,* CHAITANYA PANT, VINAY PASUPULETI, DAVID D. K. ROLSTON, ANIL JAIN,,# NARAYAN DESHPANDE,** PRIYALEELA THOTA, THOMAS J. SFERRA, and ADRIAN V. HERNANDEZ *Department of Neurological Surgery, Neurological Institute, Department of Molecular Cardiology, Department of Quantitative Health Sciences, Lerner Research Institute; Department of Internal Medicine, Department of Hospital Medicine, Medicine Institute, # einformation Technology, Cleveland Clinic, Cleveland, Ohio; Department of Pediatrics, Oklahoma University Health Sciences Center, Oklahoma City, Oklahoma; Department of Internal Medicine, Geisinger Medical Center, Danville, Pennsylvania; and **Department of Pharmacology, Bhaskar Medical College, Moinabad, Andhra Pradesh, India BACKGROUND & AIMS: In the past decade, there has been a growing epidemic of Clostridium difficile infection (CDI). During this time, use of proton pump inhibitors (PPIs) has increased exponentially. We evaluated the association between PPI therapy and the risk of CDI by performing a meta-analysis. METHODS: We searched MEDLINE and 4 other databases for subject headings and text words related to CDI and PPI in articles published from 1990 to 2010. All observational studies that investigated the risk of CDI associated with PPI therapy and used CDI as an end point were considered eligible. Two investigators screened articles independently for inclusion criteria, data extraction, and quality assessment; disagreements were resolved based on consensus with a third investigator. Data were combined by means of a random-effects model and odds ratios were calculated. Subgroup and sensitivity analyses were performed based on study design and antibiotic use. RE- SULTS: Thirty studies (25 case-control and 5 cohort) reported in 29 articles met the inclusion criteria (n 202,965). PPI therapy increased the risk for CDI (odds ratio, 2.15, 95% confidence interval, 1.81 2.55), but there was significant heterogeneity in results among studies (P.00001). This association remained after subgroup and sensitivity analyses, although significant heterogeneity persisted among studies. CONCLU- SIONS: PPI therapy is associated with a 2-fold increase in risk for CDI. Because of the observational nature of the analyzed studies, we were not able to study the causes of this association. Further studies are needed to determine the mechanisms by which PPI therapy might increase risk for CDI. Keywords: Gastric Acid Suppression; Infectious Diarrhea; Systematic Analysis; Bacteria. Clostridium difficile infection (CDI) is the leading cause of nosocomial infectious diarrhea in the developed world. 1 In the past decade, health care facilities across North America and Western Europe have reported a dramatic increase in the incidence of CDI. Based on current estimates, there are between 450,000 and 750,000 cases of CDI annually in the United States alone, 2 with an estimated 3 billion dollars spent on related health care costs. 3 A major factor driving the CDI epidemic is the emergence of a hypervirulent strain of C difficile known most commonly by the pulsed-field gel electrophoresis pattern of its DNA as the North American pulse-field type 1 strain. 4 Infection with the North American pulse-field type 1 strain typically predisposes to more severe disease and higher mortality rates when compared with other C difficile strains. 5 Historically, broad-spectrum antimicrobial therapy has been the most consistently identified causative factor involved in CDI pathogenesis including the use of fluoroquinolones. 6 More recently, a possible association between CDI and the use of proton pump inhibitors (PPIs) has been suggested. Although numerous studies have investigated a possible association between gastric acid suppressive therapy and CDI, their results have been quite conflicting. 7 Given the magnitude of the prevailing CDI epidemic and the widespread use of PPI therapy, this is a question of extreme clinical significance. In this meta-analysis, our aim was to summarize the association between PPI therapy and the risk of CDI in the published literature. Methods All procedures used in this meta-analysis were consistent with Preferred Reporting Items for Systematic reviews and Meta-analyses guidelines. 8 Data Sources and Searches We performed a systematic search of the literature using the following predetermined inclusion criteria: (1) observational studies, including cross-sectional, case-control, and cohort studies that evaluated the risk of CDI associated with PPI therapy; (2) a study population that comprised adult patients ( 18 y) who received PPI therapy; (3) CDI was a study end point; and (4) date of publication between 1990 and 2010 in any language. We used 1990 as a cut-off year because the first PPI received Food and Drug Administration approval in 1989. There was no restriction on study site (in-patient and/or outpatient). We excluded studies if: (1) there was no control group (case-control studies) or an unexposed group (cohort studies) of patients; (2) PPI use data were not available for either of the study groups; or (3) data were presented based on CDI episodes and not the number of actual patients. We identified no prospective studies evaluating PPI therapy and the risk of CDI. Abbreviations used in this paper: CDI, Clostridium difficile infection; CI, confidence interval; H2RA, histamine 2 receptor antagonist; NNH, number needed to harm; OR, odds ratio; PPI, proton pump inhibitor. 2012 by the AGA Institute 1542-3565/$36.00 doi:10.1016/j.cgh.2011.09.030

226 DESHPANDE ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 10, No. 3 This search was performed in October 2010. The following databases were searched: MEDLINE (PubMed) (1990 2010), Web of Science (1990 2010), Cumulative Index to Nursing and Allied Health Literature (CINAHL) (1990 2010), Cochrane Library, and Scopus (1990 2010). Search terms included Clostridium difficile, C. diff, C. difficile, CDAD, CDI, proton pump inhibitors, proton pumps, PPI. Reference lists from included studies and several previously published reviews on C difficile and PPI therapy also were searched. The electronic search strategy of PubMed is available in Appendix 1. Study Selection A list of retrieved articles was reviewed by 2 investigators independently (V.P. and A.D.) using the earlier-detailed eligibility criteria. Any disagreement about a particular study was decided in consensus with a third investigator (C.P.). Where more than one article for a single study was found to have been published by the same investigators, we used the most relevant publication and supplemented it, if necessary, with data from the other publications. Data Extraction and Quality Assessment Two investigators (A.D. and V.P.) independently extracted data from the full text of the included studies. Any disagreements or discrepancies were resolved in consensus with a third investigator (C.P.). Study authors were contacted if the relevant information was not available for a particular study. The methodologic quality for each article was assessed by 2 investigators independently (A.D. and V.P.). Case-control and cohort studies were assessed. Each study was assessed for study quality based on the criteria proposed by the Meta-analysis of Observational Studies in Epidemiology collaboration. 9 The study quality criteria are available in Appendix 2. Data Synthesis and Analysis The primary outcome of interest under evaluation was the association between PPI therapy and the risk of developing CDI. Although there were several potential reasons for this heterogeneity, we considered 2 a priori hypotheses to explain potential variability between studies and accordingly performed subgroup analyses: (1) study design (ie, case-control vs cohort studies); and (2) percentage of antibiotic use: for each study, antibiotic use for cases and controls in case-control studies and exposed and unexposed groups in cohort studies were calculated. Next, a median for percentage of antibiotic use was calculated for all studies (ie, case-control and cohort studies). Based on this number, studies were divided into 3 groups. These were as follows: (1) studies with percentage of antibiotic use greater than the median; (2) studies with percentage of antibiotic use less than or equal to the median; and (3) studies in which antibiotic use data were unavailable. Three a priori sensitivity analyses were planned for the primary outcome: (1) case-control studies by percentage of antibiotic use and exclusion of studies in which there was no information about antibiotic use; (2) case-control studies by percentage of antibiotic use as defined in the CDI group (ie, cases) only; and (3) exclusion of studies that evaluated recurrent CDI. We computed pooled odds ratios (ORs) and 95% confidence intervals (CIs) for our primary and all subgroup analyses. Metaanalyses were not stratified by type of PPI because all types of PPI have similar efficacy and can be used interchangeably. 10 Figure 1. Study selection process.

March 2012 PPI THERAPY PREDISPOSES TO CDI 227 DerSimonian and Laird 11 random-effects models were used for all meta-analyses. We evaluated statistical heterogeneity using the Cochran chi-square (Cochran Q) and the I 2 statistic. 12 We defined significant heterogeneity as a 2 of less than 0.10 or an I 2 statistic of greater than 50%. Reporting bias was assessed by Funnel plots. We used Review Manager for all statistical analyses (RevMan, version 5.0 for Windows; The Cochrane Collaboration, Oxford, UK). Results Study Characteristics Of the 688 references identified, 109 potentially relevant citations were selected based on relevance to the study topic. After screening all the titles and abstracts, 51 articles were selected for full-text review (Figure 1). Twenty-nine articles that reported PPI therapy and CDI published between 1990 and 2010 were included in the meta-analysis. Twenty-two articles were excluded (reasons for exclusion are listed in Figure 1). Table 1 summarizes the main characteristics of the included studies. A total of 202,965 patients were included in the metaanalysis. Of the 29 articles included in quantitative synthesis, 1 article included 2 studies. 13 Therefore, the final number of studies included in our meta-analysis was 30. Five studies were cohort studies and 25 were case-control studies. All studies were performed in hospitalized patients except 3 studies that were outpatient-based (Table 1). All studies clearly identified the study population and defined the outcome and outcome assessment (Table 2). None of the studies independently assessed the outcome, or had a selective loss of patients during follow-up evaluation. Twenty-two studies identified important confounders or prognostic factors and were used for adjustment of the association between PPI use and risk of CDI. The most common confounders included age, sex, comorbidities, antibiotic use, histamine 2 receptor antagonist (H2RA) use, and the length of stay. Main Meta-Analysis No evidence of publication bias was observed in the funnel plot (Figure 2). A meta-analysis of all studies showed a significantly higher risk of CDI with PPI therapy (OR, 2.15; 95% CI, 1.81 2.55; P.00001) (Figure 3). Significant heterogeneity of effects across studies was found (I 2 87%; P.00001). Subgroup Analyses The subgroup analysis of the 25 case-control and 5 cohort studies continued to show a higher risk of CDI with PPI therapy (OR, 2.22; 95% CI, 1.82 2.70; and OR, 1.78; 95% CI, 1.41 2.25, respectively) (Figure 3). There was no difference of Table 1. Characteristics of Studies Included in the Meta-Analysis Study, year Study location Sample size, n Mean age, y (SD) Male, n (%) Study patients Study design Shah et al, 24 2000 United Kingdom 252 81.8 85 (33.7) Hospitalized inpatients Case-control Yip et al, 25 2001 Canada 54 73 (41 89) a 26 (48.1) Hospitalized inpatients Case-control Cunningham et al, 26 2003 United Kingdom 320 NA NA Hospitalized inpatients Case-control Dial et al, 13 2004 Canada 188 74.3 (12.4) 85 (45.2) Hospitalized inpatients Case-control Dial et al, 13 2004 Canada 1187 NA NA Hospitalized inpatients Cohort Al-Tureihi et al, 27 2005 United States 53 82.3 13 (24.5) Nursing home patients Case-control Dial et al, 28 2005 United Kingdom 18,392 70.8 (16) 5630 (30.6) Community acquired Case-control Loo et al, 29 2005 Canada 474 74.5 (11.9) 241 (50.8) Hospitalized inpatients Case-control Modena et al, 30 2005 United States 250 59.7 (17.2) 128 (51.2) Hospitalized inpatients Case-control Muto et al, 31 2005 United States 406 61.5 (16 95) a 210 (51.7) Hospitalized inpatients Case-control Gillis et al, 32 2006 Canada 150 NA NA Hospitalized inpatients Case-control Kazakova et al, 33 2006 United States 195 NA (30 98) a 86 (44.1) Hospitalized inpatients Case-control Lowe et al, 34 2006 Canada 13,692 78.1 (6.8) 8268 (60.4) Outpatients Case-control Yearsley et al, 35 2006 United Kingdom 308 78.9 (10.1) 131 (42.5) Hospitalized inpatients Case-control Akhtar and Shaheen, 36 2007 United States 1290 NA (18 101) a 585 (45.3) Hospitalized inpatients Case-control Beaulieu et al, 37 2007 Canada 827 65 b 494 (59.7) Medical ICU patients Cohort Branch et al, 38 2007 United States 787 66 NA Hospitalized patients Case-control Cadle et al, 39 2007 United States 140 67 (13) 138 (98.6) Hospitalized inpatients Cohort Dubberke et al, 40 2007 United States 36,086 NA 15,159 (42) Hospitalized inpatients Case-control Jayatilaka et al, 41 2007 United States 366 67.5 (21.5) 225 (61.5) Hospitalized inpatients Case-control McFarland et al, 42 2007 United States 368 NA NA Inpatients and outpatients Case-control Aseeri et al, 43 2008 United States 188 NA 82 (43.6) Hospitalized inpatients Case-control Baxter et al, 44 2008 United States 4493 68 2167 (48.2) Hospitalized inpatients Case-control Dial et al, 45 2008 Canada 9196 77.7 (6.4) 3698 (40.2) Community acquired Case-control Dalton et al, 20 2009 Canada 14,719 68.8 (17) 7007 (47.6) Hospitalized inpatients Cohort Howell et al, 15 2010 United States 101,796 56.6 (19.9) 41,802 (41.1) Hospitalized inpatients Case-control Kim et al, 46 2010 South Korea 125 67.6 (13.9) 57 (45.6) Hospitalized inpatients Case-control Kim et al, 47 2010 South Korea 42 NA NA Hospitalized inpatients Case-control Linsky et al, 48 2010 United States 1166 73.1 (13.3) 1133 (97.2) Inpatients and outpatients Cohort Linney et al, 49 2010 Canada 284 75.65 (13) 134 (47.2) Hospitalized inpatients Case-control ICU, intensive care unit; NA, not available. a Mean (range). b Median.

228 DESHPANDE ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 10, No. 3 Table 2. Study Quality Assessment Study Year Study design Study population clearly identified? Clear definition of outcome and outcome assessment? Important confounders and/prognostic factors identified? Shah et al 24 2000 Case-control Yes Yes No Yip et al 25 2001 Case-control Yes Yes No Cunningham et al 26 2003 Case-control Yes Yes Adjusted for antibiotics and cytotoxic chemotherapy Dial et al 13 2004 Case-control Yes Yes Adjusted for age, ward, class, and number of antibiotics Dial et al 13 2004 Cohort Yes Yes Adjusted for age, ward, class, and number of antibiotics Al-Tureihi et al 27 2005 Case-control Yes Yes Adjusted for age and antibiotic treatment Dial et al 28 2005 Case-control Yes Yes Adjusted for age, sex, antibiotics, and other medications Loo et al 29 2005 Case-control Yes Yes No Modena et al 30 2005 Case-control Yes Yes Adjusted for antibiotic use and infections Muto et al 31 2005 Case-control Yes Yes Adjusted for age, diabetes, organ transplantation, H2RA, and antibiotics Gillis et al 32 2006 Case-control Yes Yes Adjusted for age, antibiotic class/number, and inpatient ward Kazakova et al 33 2006 Case-control Yes Yes Adjusted for antibiotics, H2RA, length of stay, COPD, psychosis, and depression Lowe et al 34 2006 Case-control Yes Yes Adjusted for antibiotics, other medications, and comorbidities Yearsley et al 35 2006 Case-control Yes Yes Adjusted for antibiotics Akhtar and Shaheen 36 2007 Case-control Yes Yes Adjusted for age, sex, comorbidity, and use of antibiotic or chemotherapy Beaulieu et al 37 2007 Cohort Yes Yes Adjusted for age, sex, length of stay, comorbidities, APACHE score, NGT feeding, tracheal tube placement, H2RA, and antibiotics Branch et al 38 2007 Case-control Yes Yes No Cadle et al 39 2007 Cohort Yes Yes No Dubberke et al 40 2007 Case-control Yes Yes Adjusted for age, admissions, antibiotics, CDAD pressure, albumin level, leukemia/lymphoma, mechanical ventilation, H2RA, and antimotility agents Jayatilaka et al 41 2007 Case-control Yes Yes Adjusted for antibiotics and gastric acid reducing drugs McFarland et al 42 2007 Case-control Yes Yes No Aseeri et al 43 2008 Case-control Yes Yes Adjusted for admission date, sex, age group, antibiotic use, patient location, and room type Baxter et al 44 2008 Case-control Yes Yes Adjusted for antibiotics Dial et al 45 2008 Case-control Yes Yes Adjusted for age, sex, antibiotics, H2RA, comorbidities, physician visits, hospital admissions, and length of stay Dalton et al 20 2009 Cohort Yes Yes Adjusted for number of medication groups, antibiotic days, age, length of stay, medical service, and PPI days Howell et al 15 2010 Case-control Yes Yes Adjusted for age, comorbidities, H2RA, and antibiotics Kim et al 46 2010 Case-control Yes Yes Adjusted for age, serum albumin level, and NGT feeding Kim et al 47 2010 Case-control Yes Yes No Linsky et al 48 2010 Cohort Yes Yes Adjusted for age, CDI treatment, other antibiotics, length of stay, other comorbidities, and systemic steroids Linney et al 49 2010 Case-control Yes Yes Age, sex, discharge date and hospital unit, antibiotics, diabetes mellitus, IBD, cancer, enteral feeding, length of stay, and previous residence APACHE, Acute Physiology and Chronic Health Evaluation; CDAD, C. difficile associated diarrhea; COPD, chronic obstructive pulmonary disease; IBD, inflammatory bowel disease; NGT, naso gastric tube.

March 2012 PPI THERAPY PREDISPOSES TO CDI 229 Figure 2. Funnel plot to assess publication bias. effects between case-control and cohort studies (P.1). Significant heterogeneity remained only for the case-control studies (I 2 89%). All case-control studies were further divided into 3 subgroups based on percentage of antibiotic use. Each of the 3 subgroups (ie, studies with antibiotic use greater than 80%, antibiotic use less than or equal to 80%, and antibiotic use information unavailable), showed a higher risk of CDI with PPI therapy (OR, 2.07; 95% CI, 1.50 2.86; OR, 2.73; 95% CI, 1.88 3.97; and OR, 1.77; 95% CI, 1.26 2.49, respectively) with significant heterogeneity between the studies except for studies in which antibiotic use was unavailable (Figure 4A). Sensitivity Analyses When case-control studies lacking information on antibiotic use were excluded, the association between PPI therapy and risk of CDI remained significant (OR, 2.35; 95% CI, 1.85 2.99), with significant heterogeneity between the studies (Figure 4B). Next, case-control studies were divided based on the percentage of antibiotic use as defined in the CDI group (ie, cases only). Both groups of studies with antibiotic use greater than 80% and 80% or less continued to show a higher risk of CDI with PPI use (OR, 2.46; 95% CI, 1.26 4.78; OR, 2.28; 95% CI, 1.79 2.89, respectively), with significant heterogeneity between the studies (Figure 4C). Finally, when studies that evaluated recurrent CDI were excluded from the main analysis, the association between PPI use and risk of CDI remained significant (OR, 2.13; 95% CI, 1.77 2.55), with significant heterogeneity between the studies (Figure 4D). Discussion We found an association between CDI and PPI therapy in our meta-analysis involving 30 studies and 202,965 patients. Overall, patients on PPI therapy have approximately twice the risk of developing CDI compared with nonusers. However, the significant heterogeneity amongst the studies implies that the summary OR should be interpreted with caution. On subgroup and sensitivity analyses, the positive association between PPI therapy and the risk of CDI remained; significant heterogeneity persisted between studies except in the cohort subgroup and for studies in which antibiotic use data were unavailable. The biological plausibility for an association between PPI therapy and CDI appears strong. The loss of acidity permits the proliferation of numerous species of bacteria in a previously sterile stomach; bacterial growth depends on both the duration and absolute increase in gastric ph. 14 It recently was suggested that increasing levels of pharmacologic acid suppression may be associated with increased risks of nosocomial CDI. 15 As the level of acid suppression increased, the risk of nosocomial CDI increased from an OR of 1.0 (no acid suppression) to 1.53 (95% CI, 1.12 2.10; H2RA), to 1.74 (95% CI, 1.39 2.18; daily PPI), and to 2.36 (95% CI, 1.79 3.11; more frequent PPI). The link between gastric-acid suppression and a predisposition for developing CDI was first suggested in 1993. Walker et al 16 examined C difficile colonization among residents of a long-term care facility and identified the use of H2RAs as a potential risk factor (relative risk, 3.27; 95% CI, 1.07 9.93; P.038). Since then, numerous other observational studies have investigated this possible association, for the most part with inconsistent and conflicting results. 7 In 2007, a systematic review evaluated the risk of enteric infection in patients taking acid suppression. 17 In the review, the authors identified 11 articles with a total of 126,999 patients. There was an increased risk of CDI with PPI therapy (OR, 2.05; 95% CI, 1.47 2.85), with significant heterogeneity between the studies ( 2 50.9; P.0001). More recently, 2 meta-analyses published in abstract form have evaluated PPI therapy and the risk of developing CDI. Janarthanan et al 18 combined 21 observational studies with a total of 133,054 patients in their meta-analysis. Their results indicated an 80% increase in the relative risk of CDI among PPI

230 DESHPANDE ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 10, No. 3 Figure 3. Overall analysis stratified by study design (all studies). users. The risk estimate was 1.55 for case-control studies and 2.07 for cohort studies (P.001). This study did not include subgroup analysis controlling for possible confounders such as antibiotic use. In a second study, Shukla et al 19 identified 9 case-control studies with a total of 4100 cases and 29,092 controls. Their results indicated that patients with CDI had an OR for PPI use of 2.74 compared with the controls (95% CI, 1.85 4.07; P.0001). There was significant heterogeneity among the 9 studies (P.0001). This was a relatively small meta-analysis because the only studies that adjusted for antibiotic use were included. Our meta-analysis was a large and comprehensive study that evaluated the association between PPI therapy and the risk of CDI. A major strength of our study was that we also performed subgroup analysis based on antibiotic use. Because broad-spectrum antibiotics are one of the primary risk factors that predispose patients to CDI, we believe that it is imperative to account for antibiotic use in exploratory analysis. Because of the observational nature of the analyzed studies, it is not possible to establish causality based on the current data. The sources for the considerable heterogeneity between studies included in this meta-analysis are several, but the contribution of each factor is unknown. This heterogeneity has been encountered in prior related systematic reviews as well and it has been explained by variations in the

March 2012 PPI THERAPY PREDISPOSES TO CDI 231 Figure 4. (A) Subgroup analysis of case-control studies by percentage of antibiotic use as defined in all the population (CDI no CDI). (B) Subgroup analysis of case-control studies by percentage of antibiotic use with studies with no information for antibiotic use excluded. (C) Subgroup analysis of case-control studies by percentage of antibiotic use as defined in the CDI group (ie, cases only). (D) Subgroup analysis excluding 4 studies that evaluated recurrent CDI.

232 DESHPANDE ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 10, No. 3 bacterial strain, type, dose and duration of acid-suppression therapy, concomitant illness, and medication use. 17 The reported data on the majority of these variations were insufficient to permit subgroup analysis except for the type of study and percentage of antibiotic use. In view of the considerable heterogeneity among the included studies, by using random-effects models, greater reliance could have been placed on smaller studies with potentially inferior data compared with a fixed-effects model. It should be noted, however, that although this is of particular concern for randomized controlled trials, the relationship between sample size and data quality may be less stringent for observational studies. There is a possibility of confounding by indication, that is, patients who receive PPIs and those who develop C difficile both tend to be sicker at baseline. In the United States, PPIs are freely available as over-the-counter medications. They also are heavily overprescribed by physicians; up to half of these prescriptions are for conditions that do not meet currently accepted indications for their use. How concerned should a physician be with prescribing PPI therapy for a valid indication? To answer this question, we refer to a 2009 retrospective cohort study involving 14,719 patients. 20 The results of this study indicated that PPI co-exposure increased CDI (1.44 cases/100 patients vs 0.74 cases/100 nonexposed: OR, 1.96; 95% CI, 1.42 2.72). PPI days were associated with a risk of 1.01 per day in the multivariate analysis and the average duration of PPI days per patient of 19 days indicated an average adjusted OR of 1.20. There was an excess rate of CDI of 1.8 cases per 1000, or a number needed to harm (NNH) of 545. To place this into clinical perspective, the numbers needed to treat to prevent erosive esophagitis are 2.0 and 2.5 21 and for the prevention of nonsteroidal antiinflammatory drug induced peptic ulcer disease in high-risk patients are between 7.0 and 14.0. 22 Based on data included in our meta-analysis, the calculated NNH ranged from 2 to 261 for case-control studies and from 4 to 143 for cohort studies. Most of the included studies in the meta-analysis have taken place in an epidemic setting but given the heterogeneity between baseline CDI rates in various study settings it is not justifiable to have one overall NNH for this metaanalysis. The results from our study indicate that patients on PPI therapy have a moderate risk of developing CDI compared with nonusers; therefore, we suggest physicians should not refrain from using PPI therapy for medical indications clearly established in the literature. Well-researched guidelines for the appropriate use of PPI therapy are available. 23 The use of PPIs, both prescription and over the counter, has surged in recent years. Therefore, even a moderate risk of developing CDI in PPI users, as identified in our study, translates into a significant health care problem and should serve as a statement to curb the indiscriminate use of these medications. Appendix 1. Literature Search of the PubMed Database PubMed search: 86 articles (Clostridium difficile OR c. diff OR c. difficile OR CDI OR clostridium difficile associated diarrhea OR clostridium-difficile associated diarrhea OR CDAD) AND (proton pump inhibitor OR proton pump inhibitors OR proton pump inhibitor therapy OR ppi OR proton pump OR proton-pump OR gastric acid suppressive agents). Appendix 2. Study Quality Criteria 1. Clear identification of study population? 2. Clear definition of outcome and outcome assessment? 3. Important confounders and/or prognostic factors identified? 4. Independent assessment of outcome parameters (ie, ascertainment of outcomes performed by researchers other than the ones involved in the study)? 5. Selective loss during follow-up evaluation? NOTE. Each criterion was assessed as a yes or a no answer. If a study did not clearly mention any of the earlier-listed criteria, they were assessed as a no. References 1. McFarland LV. Epidemiology of infectious and iatrogenic nosocomial diarrhea in a cohort of general medicine patients. Am J Infect Contr 1995;23:295 305. 2. Zilberberg MD, Shorr AF, Kollef MH. Increase in adult Clostridium difficile-related hospitalizations and case-fatality rate, United States, 2000 2005. Emerg Infect Dis 2008;14:929 931. 3. Kyne L, Hamel MB, Polavaram R, et al. Health care costs and mortality associated with nosocomial diarrhea due to Clostridium difficile. Clin Infect Dis 2002;34:346 353. 4. McDonald LC, Killgore GE, Thompson A, et al. An epidemic, toxin gene-variant strain of Clostridium difficile. N Engl J Med 2005; 353:2433 2441. 5. Cohen MB. Clostridium difficile infections: emerging epidemiology and new treatments. J Pediatr Gastroenterol Nutr 2009; 48(Suppl 2):S63 S65. 6. Deshpande A, Pant C, Jain A, et al. Do fluoroquinolones predispose patients to Clostridium difficile associated disease? A review of the evidence. Curr Med Res Opin 2008;24:329 333. 7. Pant C, Madonia P, Minocha A. Does PPI therapy predispose to Clostridium difficile infection? Nat Rev Gastroenterol Hepatol 2009;6:555 557. 8. Moher D, Liberati A, Tetzlaff J, et al. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med 2009;6:e1000097. 9. Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of observational studies in epidemiology: a proposal for reporting. Metaanalysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA 2000;283:2008 2012. 10. Klok RM, Postma MJ, van Hout BA, et al. Meta-analysis: comparing the efficacy of proton pump inhibitors in short-term use. Aliment Pharmacol Ther 2003;17:1237 1245. 11. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986;7:177 188. 12. Higgins JP, Thompson SG. Quantifying heterogeneity in a metaanalysis. Stat Med 2002;21:1539 1558. 13. Dial S, Alrasadi K, Manoukian C, et al. Risk of Clostridium difficile diarrhea among hospital inpatients prescribed proton pump inhibitors: cohort and case-control studies. CMAJ 2004;171:33 8. 14. Williams C, McColl KE. Review article: proton pump inhibitors and bacterial overgrowth. Aliment Pharmacol Ther 2006;23:3 10. 15. 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. 16. Walker KJ, Gilliland SS, Vance-Bryan K, et al. Clostridium difficile colonization in residents of long-term care facilities: prevalence and risk factors. J Am Geriatr Soc 1993;41:940 946. 17. Leonard J, Marshall JK, Moayyedi P. Systematic review of the risk of enteric infection in patients taking acid suppression. Am J Gastroenterol 2007;102:2047 2056; quiz, 2057.

March 2012 PPI THERAPY PREDISPOSES TO CDI 233 18. Janarthanan S, Ditah I, Kutait A, et al. A meta-analysis of 16 observational studies on proton pump inhibitor use and risk of Clostridium difficile associated diarrhea. Am J Gastroenterol 2010;105(Suppl 1):S139. 19. Shukla S, Shukla A, Guha S, et al. Use of proton pump inhibitors and risk of Clostridium difficile-associated diarrhea: a meta-analysis. Gastroenterology 2010;138(Suppl 1):S-209. 20. Dalton BR, Lye-Maccannell T, Henderson EA, et al. Proton pump inhibitors increase significantly the risk of Clostridium difficile infection in a low-endemicity, non-outbreak hospital setting. Aliment Pharmacol Ther 2009;29:626 634. 21. Caos A, Breiter J, Perdomo C, et al. Long-term prevention of erosive or ulcerative gastro-oesophageal reflux disease relapse with rabeprazole 10 or 20 mg vs. placebo: results of a 5-year study in the United States. Aliment Pharmacol Ther 2005;22:193 202. 22. Bianchi Porro G, Lazzaroni M, Imbesi V, et al. Efficacy of pantoprazole in the prevention of peptic ulcers, induced by non-steroidal anti-inflammatory drugs: a prospective, placebo-controlled, double-blind, parallel-group study. Dig Liver Dis 2000;32:201 208. 23. NHS National Institute of Health and Clinical Excellence. Dyspepsia: managing dyspepsia in adults in primary care. North of England Dyspepsia Guideline Development Group. NICE Clinical Guidelines, No. 17, 2004. 24. Shah S, Lewis A, Leopold D, et al. Gastric acid suppression does not promote clostridial diarrhoea in the elderly. QJM 2000;93: 175 181. 25. Yip C, Loeb M, Salama S, et al. Quinolone use as a risk factor for nosocomial Clostridium difficile-associated diarrhea. Infect Control Hosp Epidemiol 2001;22:572 575. 26. Cunningham R, Dale B, Undy B, et al. Proton pump inhibitors as a risk factor for Clostridium difficile diarrhoea. J Hosp Infect 2003;54:243 245. 27. Al-Tureihi FI, Hassoun A, Wolf-Klein G, et al. Albumin, length of stay, and proton pump inhibitors: key factors in Clostridium difficile-associated disease in nursing home patients. J Am Med Dir Assoc 2005;6:105 108. 28. Dial S, Delaney JA, Barkun AN, et al. Use of gastric acid-suppressive agents and the risk of community-acquired Clostridium difficile-associated disease. JAMA 2005;294:2989 2995. 29. Loo VG, Poirier L, Miller MA, et al. A predominantly clonal multiinstitutional outbreak of Clostridium difficile-associated diarrhea with high morbidity and mortality. N Engl J Med 2005;353:2442 2449. 30. Modena S, Bearelly D, Swartz K, et al. Clostridium difficile among hospitalized patients receiving antibiotics: a case-control study. Infect Control Hosp Epidemiol 2005;26:685 690. 31. Muto CA, Pokrywka M, Shutt K, et al. A large outbreak of Clostridium difficile-associated disease with an unexpected proportion of deaths and colectomies at a teaching hospital following increased fluoroquinolone use. Infect Control Hosp Epidemiol 2005;26:273 280. 32. Gillis C, Fardy J, Hutchinson J. Proton pump inhibitors: are they a risk factor for Clostridium difficile diarrhea: a case-control study. Gastroenterology 2006;20(Suppl A):191. 33. Kazakova SV, Ware K, Baughman B, et al. A hospital outbreak of diarrhea due to an emerging epidemic strain of Clostridium difficile. Arch Intern Med 2006;166:2518 2524. 34. Lowe DO, Mamdani MM, Kopp A, et al. Proton pump inhibitors and hospitalization for Clostridium difficile-associated disease: a population-based study. Clin Infect Dis 2006;43:1272 1276. 35. Yearsley KA, Gilby LJ, Ramadas AV, et al. Proton pump inhibitor therapy is a risk factor for Clostridium difficile-associated diarrhoea. Aliment Pharmacol Ther 2006;24:613 619. 36. Akhtar AJ, Shaheen M. Increasing incidence of Clostridium difficile-associated diarrhea in African-American and Hispanic patients: association with the use of proton pump inhibitor therapy. J Natl Med Assoc 2007;99:500 504. 37. Beaulieu M, Williamson D, Pichette G, et al. Risk of Clostridium difficile-associated disease among patients receiving protonpump inhibitors in a Quebec medical intensive care unit. Infect Control Hosp Epidemiol 2007;28:1305 1307. 38. Branch K, Yahl V, Kier K, et al. Gastric acid suppression by proton pump inhibitors as an independent risk factor for Clostridium difficile associated diarrhea. Pharmacol Ther 2007;32:4. 39. Cadle RM, Mansouri MD, Logan N, et al. Association of protonpump inhibitors with outcomes in Clostridium difficile colitis. Am J Health Syst Pharm 2007;64:2359 2363. 40. Dubberke ER, Reske KA, Yan Y, et al. Clostridium difficile associated disease in a setting of endemicity: identification of novel risk factors. Clin Infect Dis 2007;45:1543 1549. 41. Jayatilaka S, Shakov R, Eddi R, 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. 42. McFarland LV, Clarridge JE, Beneda HW, et al. Fluoroquinolone use and risk factors for Clostridium difficile-associated disease within a Veterans Administration health care system. Clin Infect Dis 2007;45:1141 1151. 43. Aseeri M, Schroeder T, Kramer J, et al. Gastric acid suppression by proton pump inhibitors as a risk factor for Clostridium difficileassociated diarrhea in hospitalized patients. Am J Gastroenterol 2008;103:2308 2313. 44. Baxter R, Ray GT, Fireman BH. Case-control study of antibiotic use and subsequent Clostridium difficile-associated diarrhea in hospitalized patients. Infect Control Hosp Epidemiol 2008;29: 44 50. 45. Dial S, Kezouh A, Dascal A, et al. Patterns of antibiotic use and risk of hospital admission because of Clostridium difficile infection. CMAJ 2008;179:767 772. 46. Kim JW, Lee KL, Jeong JB, et al. Proton pump inhibitors as a risk factor for recurrence of Clostridium-difficile-associated diarrhea. World J Gastroenterol 2010;16:3573 3577. 47. Kim YG, Jang BI, Choi EJ, et al. Association of proton pump inhibitors with recurrent Clostridium difficile associated disease: a matched case-control analysis by using propensity score (abstr). Gastroenterology 2010;138(Suppl 1):S-209. 48. Linsky A, Gupta K, Lawler EV, et al. Proton pump inhibitors and risk for recurrent Clostridium difficile infection. Arch Intern Med 2010;170:772 778. 49. Linney S, Fernandes T, Einarson T, et al. Association between use of proton pump inhibitors and a Clostridium difficile associated disease outbreak: case control study. Can J Hosp Pharm 2010;63:8. Reprint requests Address requests for reprints to: Abhishek Deshpande, MD, PhD, Department of Neurological Surgery, Neurological Institute, 9500 Euclid Avenue, Desk S60, Cleveland Clinic, Ohio 44195. e-mail: deshpaa@ccf.org; fax: (216) 445-6878. Acknowledgments A.D., C.P., and V.P. contributed equally to this work. Conflicts of interest The authors disclose no conflicts.