Prevalence of Clostridium difficile Infection among Solid Organ Transplant Recipients: A Meta-Analysis of Published Studies

Several factors including antibiotic use, immunosuppression and frequent hospitalizations make solid organ transplant (SOT) recipients vulnerable to Clostridium difficile infection (CDI). We conducted a meta-analysis of published studies from 1991-2014 to estimate the prevalence of CDI in this patient population. We searched PubMed, EMBASE and Google Scholar databases. Among the 75,940 retrieved citations, we found 30 studies coded from 35 articles that were relevant to our study. Based on these studies, we estimated the prevalence of CDI among 21,683 patients who underwent transplantation of kidney, liver, lungs, heart, pancreas, intestine or more than one organ and stratified each study based on the type of transplanted organ, place of the study conduction, and size of patient population. The overall estimated prevalence in SOT recipients was 7.4% [95%CI, (5.6-9.5%)] and it varied based on the type of organ transplant. The prevalence was 12.7% [95%CI, (6.4%-20.9%)] among patients who underwent transplantation for more than one organ. The prevalence among other SOT recipients was: lung 10.8% [95% CI, (5.5%-17.7%)], liver 9.1 % [95%CI, (5.8%-13.2%)], intestine 8% [95% CI, (2.6%-15.9%)], heart 5.2% [95%CI, (1.8%-10.2%)], kidney 4.7% [95% CI, (2.6%-7.3%)], and pancreas 3.2% [95% CI, (0.5%-7.9%)]. Among the studies that reported relevant data, the estimated prevalence of severe CDI was 5.3% [95% CI (2.3%-9.3%)] and the overall recurrence rate was 19.7% [95% CI, (13.7%-26.6%)]. In summary, CDI is a significant complication after SOT and preventive strategies are important in order to reduce the CDI related morbidity and mortality.


Introduction
Clostridium difficile is the most common cause of hospital-acquired infections [1]. Recent studies report the prevalence of Clostridium difficile infection (CDI) among hospitalized patients to be 0.9% [2]. Studies have reported an increase in hospitalization rates associated with C. difficile infection (CDI) [3]. There has also been a remarkable increase in mortality among CDI patients in hospitals [4], as mortality from CDI increased five-fold from 1999/2000-2005/2006 [5] and CDI accounts for! $4.8 billion in excess health-care costs [6].
Solid organ transplant (SOT) recipients are at high risk for CDI because of impaired defense mechanisms resulting from immunosuppression, perioperative antibiotic use and organ failure [7][8][9]. A study among 49,198 SOT recipients that used data from the 2009 US inpatient sample database noted that these patients are at greater risk for CDI compared to the general hospital population and estimated the prevalence of infection in SOT patients to be 2.7% [10]. However, this earlier study captured CDI episodes that occurred in an indeterminate time after transplantation and did not provide data for the initial hospitalization period post-transplantation where the most CDI episodes are expected to occur. In order to address these issues, we performed a meta-analysis to estimate the prevalence of CDI in SOT patients during the peritransplant and post-transplantation period in university-based, tertiary medical centers. Also, we aimed to stratify the results based on type of the organ transplanted and estimate the recurrence rate and severity of CDI in this specific patient population.

Study Selection
We (S.P and I.M.Z) searched PubMed (1978 to February 2015), EMBASE and Google Scholar databases to identify studies that reported the prevalence of CDI among SOT recipients. The concise search term for PubMed was transplant Ã AND (clostrid Ã OR difficile OR diarrhea OR infect Ã OR (clostridium difficile) OR (pseudomembranous colitis)). The terms infect Ã and diarrhea were included in the search term in order to retrieve all articles that report episodes of CDI along with other infections, as well as episodes of CDI along with other causes of diarrhea in SOT patients. Articles that were considered eligible by title and abstract reading were assessed in full text. The reference lists of the eligible studies were also reviewed to find possible studies that match our search. Our meta-analysis follows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (S1 Table) [11].

Inclusion Criteria
Only studies that reported the prevalence of CDI among SOT patients during the peri-transplant period were included. Peri-transplant period was defined as "the time of transplant to the first discharge from the hospital" [12]. Studies that did not include follow-up of patients during the initial hospitalization post transplantation were excluded. Studies with adequate quality as described below in the Quality Assessment section were included. Also, studies published in a language other than English were excluded from the analysis.

Outcomes of Interest
The primary outcome of interest of this meta-analysis was the prevalence of CDI among SOT patients. CDI was defined as "the presence of symptoms (usually diarrhea) and either a stool test result positive for C. difficile toxins or toxigenic C. difficile or colonoscopic findings demonstrating pseudomembranous colitis" [13]. Prevalence was calculated as the proportion of the patients diagnosed with CDI among the patients "at risk", i.e. patients who received solid organ transplantation. A subgroup analysis was performed according to the type of organ transplanted, location of study conduction, and size of study population.

Data Extraction
Studies that were considered for inclusion in the meta-analysis were evaluated by two reviewers (S.P. and I.M.Z.) and all relevant information from the text, figures, tables and charts were extracted for analysis. Studies that contained duplicate information were included only once. Extracted data include period of the study, patient population and location. The total number of patients who underwent SOT during the study period and total number of CDI cases among them were also extracted. The median follow-up duration, methods of C. difficile isolation, study design, number of severe cases, and number of recurrent episodes were also included. Recurrent CDI episodes were defined as onset of symptoms after complete abatement of symptoms with proper antibiotic therapy with additional positive CDI assays. Only data about the first CDI recurrence were used for the estimation of the recurrent rate in our study population. Severe cases were represented by cases that needed surgery for colitis and/or cases that required admission in the intensive care unit due to complications directly related to CDI and/or died from CDI-related cause [10].

Quality Assessment
The methodological quality of eligible studies was assessed by two reviewers (S.P. and I.M.Z.) using the Newcastle-Ottawa Quality Assessment Scale, which is a 'star based' rating system [14]. The parameters used to evaluate the quality of individual studies were representativeness of the exposed cohort, ascertainment of exposure, demonstration that outcome of interest was not present at start of the study, assessment of outcome, follow-up long enough for outcomes to occur and adequacy of follow up of cohorts [15]. Two parameters 'selection of the non-exposed cohort' and 'comparability of cohorts on the basis of the design or analysis' were not applicable to our analysis, so each article could get up to 6 stars. We considered the study population representative of the exposed cohort if data on CDI were provided for all available transplant patients and not among a specific sub population. The outcome was assessed by the cases who presented with symptoms and laboratory diagnosis of CDI. Follow-up duration of at least 3 months was considered adequate for the outcome to occur in the cohort. Studies that received 5 stars were considered of adequate quality for extraction of relevant information.

Statistical Analysis
We performed the meta-analysis using a random-effects model to estimate the pooled prevalence and the 95% confidence intervals (CI), using Der-Simonian and Laird weights [16]. The double arcsine methodology was used to avoid an undue large weight for studies with low or high prevalence (prevalence close to 0 or 1) [17]. The Egger's test [18,19] was used to assess small study effects. Between-study variance τ 2 estimation [20] was used to assess statistical heterogeneity. Subgroup analyses were used to account for possible sources of heterogeneity. The meta and metareg commands of the Stata v13 software package (Stata Corporation, College Station, TX) were used to perform the statistical analysis. The statistical significance threshold was set at 0.05.

Results
Our literature search yielded 75,940 citations and the last day of the literature search was February 9, 2015. After scrutinizing the titles and abstracts of retrieved articles, 162 articles were accessed in full text. Among these 162 articles, 127 studies were excluded because they did not contain extractable data on prevalence of CDI among SOT patients. Of the total 35 remaining eligible studies, 10 contained partially overlapping data [21][22][23][24][25][26][27][28][29][30] and data were extracted with an effort to extract the maximum available data. We included 30 studies in the final analysis coded from 35 articles. The details of the selection process of eligible articles are presented in the Flow chart (Fig 1).
Geographical location varied among the included articles. Among the 30 studies, 22 were conducted in the North America, 7 in Europe, and 1 in Asia. All studies were retrospective. Twenty one studies reported the method of isolation of C. difficile. One or more than 1 method of isolation such as stool culture and toxin detection were used in the studies. C. difficile toxin was identified using one or more of the following methods: enzyme immunoassay, radio immunoassay, tissue culture assay and PCR. The method of C. difficile isolation used in each individual study is included in Table 2.
The pooled prevalence of CDI among 21,683 SOT recipients was 7.4% [95%CI, (5.6-9.5%), τ 2 = 0.039] (Fig 2). According to the Egger's test, we found no evidence of small study effects for the overall estimated prevalence (bias: 1.93, p-value: 0.06). We also stratified our data based on the type of organ transplanted. The prevalence of CDI inpatients who underwent transplantation for more than one organ was 12. The differences in prevalence between kidney and liver (p-value: 0.07) and kidney and lung (p-value: 0.07) were marginally significant. The prevalence of CDI among kidney transplant recipients was also lower than intestine, heart and multiple organ recipients, but these differences did not reach statistical significance [kidney-heart (p-value: 0.83), kidney-intestine (p-value: 0.48) and kidney-multiple organs (p-value: 0.13)].
Fifteen studies reported data on recurrence of CDI among 1,020 infected patients. Among them, 10 studies included the definition of recurrence. Three studies defined an episode as recurrent when the diagnosis was done within 8 weeks from the initial episode, 2 had a time frame longer than 8 weeks and the remaining 5 did not specify the duration (Table 2). Those studies that used a longer follow-up period to define recurrent episodes tend to overestimate the rate of recurrence. The reported recurrence rate was estimated to be 19.7% [95% CI, (13.7%-26.6%), τ 2 = 0.073]. Definition of recurrence, time to recurrence, mean time to diagnosis and methods of isolation of C. difficile for individual studies are presented in the Table 2. Finally the reported prevalence of severe CDI from 19 studies was 5.3% [95% CI (2.3%-9.3%), τ 2 = 0.097] and among the 1,243 C. difficile infected patients that were included in these studies there were 27 reported cases of CDI related colectomy, 37 patients had ICU admission and 20

Discussion
C. difficile infection has been increasingly recognized among SOT recipients [31]. The estimated prevalence of CDI in SOT patients in our study was 7.4%, which is higher than the 0.9% reported in the general hospital population [2]. In our study, lung and liver transplant patients had higher prevalence of CDI compared to kidney transplant recipients and this difference was marginally significant. There was no significant difference in the prevalence based on the size of the study population. Recurrence of CDI episodes was seen in 19.7% of cases and 5.3% of patients had severe CDI resulting in colectomy, ICU admission or death. The overall estimated prevalence of CDI in our study is almost 3 times higher than that reported in a previous study which used data from 2009 nationwide inpatient sample database (2.7%) [10]. The fact that we included data from the initial hospitalization after transplantation might be the reason for the higher overall estimated prevalence. Indeed, most cases of CDI among SOT recipients are diagnosed in the early post-transplantation period due to intense immunosuppression, more frequent antimicrobial exposure and increased exposure to the health care setting [7]. The observed difference might also be due to the long duration of follow-up in the studies included in our meta-analysis as the follow-up time in most of our studies was up to several months and even years after transplantation.
In agreement to the study that reported the data from nationwide inpatient sample database [10], the prevalence of CDI was higher among lung transplant recipients in comparison to the kidney transplant patients. This might be attributed to the higher level of immunosuppression required for lung transplant patients [24]. Also, lung transplant patients often have a history of prolonged exposure to antibiotics and frequent hospital admissions, increasing their risk for CDI [25]. Also, patients who underwent transplantation of more than one organ had more than 3-times higher point prevalence of CDI than kidney recipients in our study, but this difference did not reach statistical significance, possibly due to the limited number of cases of multi-organ transplant patients.
Recurrence is a common problem in SOT patients with CDI [32]. Recurrent CDI can occur either because of relapse or because of re-infection; the relative frequency of each of these mechanism has not been well described [33]. Risk factors, such as increased length of hospital stay, prolonged use of antibiotics, immunosuppression and co-morbid conditions are associated with recurrence of CDI in SOT recipients [31,34] and, the recurrence rate varies based on the treatment of the initial CDI episode [35]. The estimated recurrence rate of CDI in our study was 19.7%, and it was comparable to the median 21.6% that was recently reported in a relevant meta-analysis in the general hospital setting [36].
Knowledge on the outcomes of CDI in SOT recipient is remarkably limited [37]. Few studies have reported an increase in the in hospital mortality, a longer hospital stay and cost of health care services among this group of patients [10,38]. CDI can also result in complications requiring colectomy and ICU admission [10]. Among the studies that reported data on severity, at least 5.3% of the patients had complications related to CDI in compared to 3%-5% rate of fulminant colitis in total hospitalized C. difficile infected patients [9,39,40]. The prevalence of colectomy in SOT patients due to CDI related complications in our study was 2.7% compared with 0.7% reported from the national inpatient sample database analysis in the general hospital population from 2001-2010 [41]. To locate possible sources of heterogeneity, we performed a subgroup analysis based on the size of the study population. Studies with small population size tend to overestimate the outcome. To evaluate this effect we stratified the studies based on population size (!200 versus <200). The size of study population, however, did not significantly alter the estimated prevalence of CDI in SOT recipients (Table 3).
Notably, there were certain limitations in our study. Most of the studies that were included in our meta-analysis had a long study period (>6 years) and did not report the prevalence of CDI stratified by year. Therefore, it was not feasible to estimate the trend over time, so it is likely that the number of CDI cases in the recent times can be even higher. The follow-up period varied widely among studies included in our analysis making a sub-analysis of the data based on follow-up duration not possible. However, out of 13 studies that reported the mean time to CDI diagnosis, only 3 had a mean time of more than 3 months with the remaining 10 studies having a mean time to diagnosis within 3 months, supporting that most of the recorded episodes of CDI occurred shortly after SOT. Detection methods used to diagnose CDI among the studies included in our analysis differ in sensitivity [42]. Two or more methods of isolation of C. difficile were used in most of the studies and during different time frames, but the studies did not provide stratified extractable data. Therefore, a sensitivity analysis based on the diagnostic methods used was not feasible. Also, the definition of recurrence varied between studies [24,31,38,43] and in 2 of them the time to relapse of symptoms used for defining recurrent cases was longer than the 8 weeks used in current clinical practice guidelines for Clostridium difficile infection in adults [44]. The latter studies tend to overestimate the number of recurrent cases and therefore the estimated prevalence of recurrence in our study might not be a precise estimate as per the current clinical practice guidelines. The estimated recurrence rate in our study is a pooled data from studies using different treatment modalities, so different rate of treatment failure and recurrence might exist in each individual study. Twenty studies in our analysis mentioned possible risk factors associated with CDI in SOT recipients. They described prolonged exposure to antibiotics [27,43,[45][46][47][48][49][50], immunosuppression [21,46,[48][49][50][51] and increased days of hospitalization [26,27,38,43,45,46,50,52,53] to be associated with CDI. However, due to lack of raw data on each risk factor, statistical representation could not be performed in our study. Importantly, infection control measures and local epidemiology have a significant role in the prevalence of CDI among individual centers and our pooled estimation does not decrease the need of each local center to know the local prevalence.
In conclusion, our analysis estimated the pooled prevalence of CDI among SOT recipients to be almost 3 times higher than previously indicated. The observed high prevalence of CDI, along with the significant rate of severe cases, highlights the need for preventive policies, such as antimicrobial stewardship programs, strict compliance with hand hygiene and environmental decontamination that specifically target this patient population. Also, studies are needed to identify immunosuppressive and prophylactic antimicrobial regimens that are probably associated with lower risk of CDI.