The Association between Histamine 2 Receptor Antagonist Use and Clostridium difficile Infection: A Systematic Review and Meta-analysis

Background Clostridium difficile infection (CDI) is a major health problem. Epidemiological evidence suggests that there is an association between acid suppression therapy and development of CDI. Purpose We sought to systematically review the literature that examined the association between histamine 2 receptor antagonists (H2RAs) and CDI. Data source We searched Medline, Current Contents, Embase, ISI Web of Science and Elsevier Scopus from 1990 to 2012 for all analytical studies that examined the association between H2RAs and CDI. Study selection Two authors independently reviewed the studies for eligibility. Data extraction Data about studies characteristics, adjusted effect estimates and quality were extracted. Data synthesis Thirty-five observations from 33 eligible studies that included 201834 participants were analyzed. Studies were performed in 6 countries and nine of them were multicenter. Most studies did not specify the type or duration of H2RAs therapy. The pooled effect estimate was 1.44, 95% CI (1.22–1.7), I2 = 70.5%. This association was consistent across different subgroups (by study design and country) and there was no evidence of publication bias. The pooled effect estimate for high quality studies was 1.39 (1.15–1.68), I2 = 72.3%. Meta-regression analysis of 10 study-level variables did not identify sources of heterogeneity. In a speculative analysis, the number needed to harm (NNH) with H2RAs at 14 days after hospital admission in patients receiving antibiotics or not was 58, 95% CI (37, 115) and 425, 95% CI (267, 848), respectively. For the general population, the NNH at 1 year was 4549, 95% CI (2860, 9097). Conclusion In this rigorous systematic review and meta-analysis, we observed an association between H2RAs and CDI. The absolute risk of CDI associated with H2RAs is highest in hospitalized patients receiving antibiotics.


Introduction
Clostridium difficile infection (CDI) is considered a major health problem with a point prevalence of 13.1/1000 in-patient [1] and is increasing in incidence and mortality [2][3][4][5]. The CDI cost in the United States of America (USA) alone was conservatively estimated to exceed $1.1 billion annually [6]. Risk factors associated with CDI acquisition are numerous and traditionally have included exposure to antibiotics, advanced age, comorbidities, enteral feeding, prolonged hospitalization, endoscopy and antineoplastic medications [7][8][9][10].
The role of gastric acid suppression therapy has gained interest recently as a risk factor for CDI. Four recently published metaanalyses have suggested an association between gastric acid suppression therapy with proton pump inhibitors (PPI) and CDI [11][12][13][14]. The United States Food and Drug Administration (FDA) recently warned the public about a possible association between CDI and PPI use [15]. However, to date; there is no systematic review dedicated to evaluate the potential association between histamine 2 receptors antagonists (H 2 RAs) use and risk of CDI.
H 2 RAs are popular over-the-counter (OTC) drugs worldwide [16]. Off -label use of H 2 RAs and substitution for physician care were reported in 46 % and 34% of the adult consumer, respectively [15]. Masking serious conditions, missed diagnosis, and the potential for inappropriate use by patients are concerns about OTC use of H 2 RAs [17]. Nonetheless, the implications of OTC H 2 RAs use are not yet well defined.
Given the high prevalence of prescription use and OTC use of H 2 RAs and the increasing incidence and severity of CDI, we sought to systematically review the published literature that examined the association between H 2 RAs use and development of CDI following the MOOSE [18] and PRISMA [19] guidelines. We use the Grades of Recommendation, Assessment, Development and Evaluation (GRADE) framework [20] to interpret our findings.

Search strategy
The search strategy and subsequent literature searches were performed by a medical reference librarian (PJE) with 37 years of experience. The initial strategy was developed in Ovid MEDLINE (1990 through January 2012), using MeSH (Medical Subject Headings) controlled vocabulary, and then modified for Ovid EMBASE (1990 through January 2012). Primary terms were: enterocolitis, pseudomembranous/ AND the therapeutic agents of interest: explode omeprazole, explode proton pump inhibitors, anti-ulcer agents, and explode histamine H 2 antagonists (Explode allows including all of the specific drugs, without having to use all of the various terms, synonyms, brands and generic names.) Articles were limited to randomized controlled trials, cohort studies, and or case-control studies. The same process was used with Ovid EMBASE with alterations as necessary to accommodate EMBASE's more granular subject headings. ISI Web of Science and Elsevier Scopus use text words: (difficile OR pseudomembranous OR pseudo-membranous) AND (omeprazole OR ''proton pump'' OR ranitidine OR h2 OR h-2 OR ''acid suppression'' OR antacid*)) AND (random* OR trial* OR blind* OR cohort* OR controlled OR prospective). Moreover, bibliographic references of

Study selection
To be included, a study had to: (1) be an analytical study; and (2) examine the association between H 2 RAs use and incidence of CDI in adult population.

Data collection
A data collection form was developed and used to retrieve information on relevant features and results of pertinent studies. Two reviewers (A.B.A. and F.A.) independently extracted and recorded data in a predefined checklist. Disagreements among reviewers were discussed with two other reviewers (I.M.T. and M.A.A.), and agreement was reached by consensus. We collected adjusted effect estimates and 95% confidence intervals (CI) based on the multivariable regression model used in each study.
We used the Newcastle-Ottawa Quality Assessment Scale for cohort and case-control studies [21] which is intended to rate selection bias, comparability of the exposed and unexposed groups of each cohort, outcome assessment, and attrition bias. Two reviewers (M.A.G and F.A.) independently assessed the methodological quality of selected. Disagreement among reviewers was discussed with 2 other reviewers (I.M.T. and M.A.A.), and agreement was reached by consensus.
We used the GRADE framework to interpret our findings. The Cochrane Collaboration has adopted the principles of the GRADE system [20] for evaluating the quality of evidence for outcomes reported in systematic reviews.
For purposes of systematic reviews, the GRADE approach defines the quality of a body of evidence as the extent to which one can be confident that an estimate of effect or association is close to the quantity of specific interest. Quality of a body of evidence involves consideration of within-study risk of bias (methodological quality), directness of evidence, heterogeneity, precision of effect estimates and risk of publication bias.

Statistical Analyses
Meta-analyses. The primary effect measures used in the meta-analysis were Odds Ratios (OR), Hazard Ratios (HR) and Relative Risks (RR) which were assumed to reasonably estimate the same association between CDI and H 2 RAs given the low   incidence of CDI and thus were pooled together. Adjusted effect estimates were primarily used for this analysis. Unadjusted effect estimates were used as alternatives if studies did not pursue adjustment because of absence of association on univariate comparison. Effect estimates from all included studies were pooled in a metaanalysis weighing individual studies according to their logtransformed inverse variance. The DerSimonian and Laird random effects model [22] was used to calculate the pooled effect estimates.
We extracted data on the proportion of CDI cases that were exposed to antibiotics from all studies that reported these data. We then performed a meta-analysis for the proportion on logit scale using random effects model weighing the individual studies according to their log-transformed inverse variance.
Exploring heterogeneity. Homogeneity among studies was tested by means of Cochran's Q test and calculation of the variation across studies attributable to heterogeneity rather than chance (I 2 ). The influence of a range of a-priori selected study-level and aggregated individual-level parameters on the observed effect    (1) Representativeness of the exposed cohort: A, truly representative; B, somewhat representative; C, selected group; D, no description of the derivation of the cohort.
(2) Selection of the non-exposed cohort: A, drawn from the same community as the exposed cohort; B, drawn from a different source; C, no description of the derivation of the non-exposed cohort.
(4) For demonstration that the outcome of interest was not present at start of study: A, yes; B, no.
N Comparability: For comparability of cohorts on the basis of the design or analysis: A, study controls for co-morbidities; B, study controls for any additional factor (e.g., age and severity of illness); C, not done. 0 Outcome: (1) Assessment of outcome: A, independent blind assessment; B, record linkage; C, self-report; D, no description.
(2) Was follow-up long enough for outcomes to occur? A, yes, (i.e. in-hospital or up to 30 d); B, no.
(3) Adequacy of follow-up of cohorts: A, complete follow-up and all subjects accounted for; B, subjects lost to follow-up was unlikely to introduce bias, because a small number were lost or a description was provided of those lost; C, follow-up rate 90% or lower (select an adequate percentage) and no description of those lost; D, no statement. doi:10.1371/journal.pone.0056498.t005 estimate was investigated by means of meta-regressions. In these analyses, the log odds ratio from each study was regressed on the potential confounders in univariate and multivariate weighted linear regressions, weighted according to the inverse standard error and the residual between-study variance. Ten potential confounders were considered. Seven variables were categorical: design of the study (case-control vs. cohort), country of publication, setting (single center vs. multicenter), method of ascertainment of antibiotic use, method of effect measure (OR vs. RR/HR), effect estimate (adjusted vs. unadjusted) and quality of included studies (high score vs. low score). Three continuous variables were: the impact factor of the journal where the study was published, number of variables the effect measure was adjusted for and proportion of cases that were exposed to antibiotics. Publication bias. The possible influence of publication bias was graphically assessed with the novel method of contourenhanced funnel plot where log-transformed odds ratios were plotted against standard errors. This method examines whether any funnel plot asymmetry is likely to be due to publication bias compared with other underlying causes of funnel plot asymmetry. The contours help to indicate whether areas of the plot, where studies are perceived to be missing, are where studies would have statistically significant effect sizes or not and thus decrease or increase the evidence that the asymmetry is due to publication bias. The presence of funnel plot asymmetry was also assessed using Egger's test [23].
Residual confounding. Finally, the possible influence of unknown confounders (residual confounding) was investigated with a rule-out approach described by Schneeweiss [24]. This approach stipulates the influence of a hypothetical confounder and determines what characteristics this confounder must have to fully account for the observed association between use of H 2 RAs and occurrence of CDI. The hypothetical confounder is characterized by its association to H 2 RAs use (OR EC , odds ratio of exposure to the confounder) and its association to the outcome (RR CO , relative risk of outcome in individuals exposed to the confounder vs. nonexposed). For this analysis, the absolute risk in the pooled nonexposed group was used for conversion of odds ratio to relative risk using the method described by Zhang and Yu [25]. Separate analyses were performed to demonstrate what levels of OR EC and RR CO would be required to fully explain the observed association between H 2 RAs and CDI for different hypothetical prevalence of the unknown confounder (i.e. P C = 0.2, P C = 0.4) before and after adjustment for publication bias as described above.
In all analyses, results associated with p-values ,0.05 (two-sided test) were considered statistically significant. All statistical analyses were performed using Stata version 12 statistical software (StataCorp, College Station, Texas).

Search results
The search yielded 27 eligible studies after excluding 260 citations. Six more studies were retrieved from recent review articles and added to the total eligible studies. Kutty [26] et al and Jayatilaka [27] et al, each reported 2 different observations for Figure 3. Forest plot of the pooled proportion of Clostridium difficile cases that were exposed to antibiotics. doi:10.1371/journal.pone.0056498.g003 Table 6. Influence of study type and country on the pooled effect estimate and its associated heterogeneity. different participants. Thus, a total of 33 articles met our inclusion criteria representing 35 observations that included 201834 participants. There was excellent agreement for the inclusion of the studies, data abstraction and quality assessment between the reviewers (kappa statistic being 1.0, 1.0 and 0.91 respectively).

Quality assessment
Quality assessment of all included studies was done using the validated Newcastle-Ottawa Quality Assessment Scale [21] for cohort and case control studies (Tables 4 and 5 Most studies were of good quality with no evidence of selection bias, and with good comparability of the exposed and unexposed groups of each cohort, and outcome assessment.

Meta-analysis
Thirty-five observations from 33 eligible studies were pooled using a random effect model meta-analysis. We excluded the study by Jenkins et al. as an outlier due to its large standard error. The pooled effect estimate was 1.44, 95% CI (1.22-1.7), I 2 = 70.5%. The pooled effect estimate for high quality studies was 1.39 (1.15-1.68), I 2 = 72.3%.
Although the heterogeneity between the analyzed studies was moderate, the majority of studies pointed towards a positive association. Figure 2 shows the forest plot and the pooled effect estimate for all studies stratified by country. Table 6 summarizes the pooled estimates and associated heterogeneity across different subgroups. The pooled proportion of CDI cases that were exposed to antibiotics was 0.81, 95% CI (0.65-0.91) as shown in Figure 3.

Exploring heterogeneity
The influence of a range of a-priori selected study-level and aggregated individual-level parameters on the observed effect estimate was investigated by means of meta-regressions. Table 7 summarizes the meta-regression analyses for all 35 results. Heterogeneity could not be explained by any of the 10 considered variables.

Residual confounding
The results of the residual confounding analysis are presented in Figure 5. Panel A refers to a confounder with a prevalence of 0.20; at this prevalence level, a strong confounder causing a two-fold increased risk of CDI would have to be severely imbalanced between H 2 blockers users and non users (OREC = 8.87) in order to fully account for the observed adjusted RR of 1.40. For a very common confounder with a prevalence of 0.40, stronger associations with acid-suppression use and/or CDI would be needed to explain the observed association between acidsuppression use and CDI. At this prevalence level, the confounder would have to be both imbalanced (OREC = 5.87) and increase the CDI risk (2.5-fold) to account for the observed OR, after taking publication bias into account.

Number needed to harm
The number needed to harm (NNH) was estimated by using the pooled OR from the meta-analysis [59]. A recent large prospective hospital cohort [48] reported the incidence of CDI at 14 days after hospital admission in patients receiving antibiotics or not: which was 42/1,000 and 5.4/1000, respectively. Based on these reported baseline risks, the number needed to harm (NNH) was 58, 95% CI (37,115) and 425, 95% CI (267, 848), respectively. For the general population, the NNH at 1 year was 4549, 95% CI (2860, 9097) at 1 year, based on a baseline incidence of CDI of 48/100,000 person-years [60].

Findings
In this rigorously conducted systematic review and metaanalysis, we observed an association between H 2 RAs use and development of CDI. Using the GRADE framework, the evidence supporting this association is considered of moderate quality. Although evidence from observational studies is considered of weak quality, we have ruled out a strong effect of an unmeasured confounder and, therefore, have upgraded its quality to moderate evidence in favor of this association.
The absolute risk of CDI was highest in hospitalized patients receiving antibiotics with an estimated NNH of 58 at 2 weeks. In contrast, the risk was very low (4549) in the general population. We also observed that, on average, 19% of CDI cases had not been recently exposed to antibiotics.
These findings add to previous subgroup analyses of a limited number of H 2 RA studies performed in a recent systematic review of the association between PPI and CDI. In this review, Kwok [11]

Biologic plausibility
The pathogenic mechanisms operative in H 2 RAs therapy causing an increased risk of CDI acquisition are unclear, because gastric acid does not kill gastric C. difficile spores. One potential explanation for the association between CDI and gastric acid suppression therapies could be that the vegetative form of C. difficile, which is killed by acid, plays a role in pathogenesis. Vegetative forms survive on surfaces and could be ingested by patients [62]. Survival of acid-sensitive vegetative forms in the stomach could be facilitated by two primary factors: (1) suppression of gastric acid production by acid-suppressive medications; and (2) presence of bile salts in gastric contents of patients on acid-suppressive therapy. Bile salts, which are mainly found in the small intestine, are present in gastric contents, particularly among patients with gastro-esophageal reflux disease (GERD).
The extent of gastric acid suppression could play an important role in potentiating the risk of infection. Kwok [11] et at compared the risk of CDI with gastric acid suppression from 15 studies that reported on estimates of both PPI and H 2 RAs independently on their sample of participants and found that PPI is associated with higher risk of infection in comparison to H 2 RAs though both increase the risk.

Implications
Our findings have global implications both on the inappropriate use of acid-suppression therapy and on the increasing incidence of CDI.
Given the relatively low NNH (58 patients) needed to cause a case of CDI in hospitalized patients receiving antibiotics it becomes necessary to judiciously use H 2 RAs in these patients. In addition, reducing the inappropriate use of acid-suppression . Contour-enhanced funnel plot of the association between the estimated effect size and its standard error in all studies comparing those exposed and unexposed to H2RA displays areas of statistical significance on a funnel plot. Contours represent conventional ''milestone'' levels of statistical significance (e.g., ,0.01, ,0.05, ,0.1). This funnel plot is symmetrical as it is not missing studies in the white area excluding the possibility of publication bias (Egger's test, p = 0.905). doi:10.1371/journal.pone.0056498.g004 medications in this patient population could lead to a significant reduction in the incidence of CDI.
On the other hand, our findings are re-assuring to the public that H 2 RAs use in the general population as over-the-counter medications do not pose significant CDI risk and is associated with a high NNH.

Strengths
Our study has several important strengths. This review is the first systematic evaluation dedicated to examine the association between H 2 RAs and risk of CDI. It includes a comprehensive, upto-date literature search and formal assessment of the methodological quality of pertinent studies with the largest number of relevant studies as compared to previous reviews [11,61]. In addition, our pooled estimates are based on multivariate ORs of studies adjusting for several important CDI risk factors. We also performed subgroup analyses and sensitivity analyses that confirmed the robustness of our main results. There was no statistical evidence of publication bias and the effect of residual confounding on the observed association was examined. Finally, the NNH in different risk groups was calculated to aid physicians and patients in making a decision to use H 2 RA or not.

Limitations
Our review has certain limitations. There was moderate between-study heterogeneity; however, this is often the case in meta-analyses of large observational studies [63][64][65]. Moreover the majority of studies pointed towards a positive association. There was virtually no qualitative heterogeneity, and subgroup and sensitivity analyses showed results consistent with the main analysis. There are many patient level parameters which may have led to substantial heterogeneity. Nevertheless, investigating these variables is only possible with individual patient data metaanalysis.

Conclusions
In this rigorous systematic review and meta-analysis, we observed an association between H 2 RAs and CDI. The absolute risk of CDI associated with H 2 RAs was highest in hospitalized patients receiving antibiotics. On the other hand, our findings are re-assuring that H 2 RAs use in the general population as over-thecounter medications do not pose a significant CDI risk.  Author Contributions Figure 5. Influence of a hypothetical dichotomous confounder present in 20% (panel A) and 40% (panel B) of the study population, unaccounted for in prior adjustments performed in individual studies. The graphs depict what combinations of OREC and RR would be necessary for the confounder to fully account for the observed association between H2RA use and CDI acquisition. Abbreviations: OR EC , odds ratio of exposure to the confounder in H2RA nonusers vs. H2RA users; RR CD , relative risk of CDAD in individuals exposed to the confounder vs. non-exposed. doi:10.1371/journal.pone.0056498.g005