Eur J Clin Microbiol Infect Dis DOI 10.1007/s10096-015-2348-3
ARTICLE
Predicting Clostridium difficile infection in diabetic patients and the effect of metformin therapy: a retrospective, case–control study N. Eliakim-Raz & G. Fishman & D. Yahav & E. Goldberg & G. Y. Stein & H. B. Zvi & A. Barsheshet & J. Bishara
Received: 7 January 2015 / Accepted: 2 February 2015 # Springer-Verlag Berlin Heidelberg 2015
Abstract Data on risk factors for Clostridium difficile infection (CDI) in diabetic patients are scarce. Recently, it has been shown that metformin increases the Bacteroidetes/Firmicutes ratio; therefore, it may yield a protective effect against CDI. We aimed to assess risk factors for CDI in diabetic patients beyond antibiotic treatment, and to determine the impact of metformin therapy on the development of CDI in these patients. In this retrospective, case–control study, all consecutive CDI diabetic patients, from January 2009 to December 2013, were included and compared to consecutive diabetic patients without CDI, hospitalized during the same period and in the same departments. Of 7,670 patients tested for C. difficile toxins, 486 were diabetics. Of them, 150 (30.8 %) were positive for C. difficile toxins and 336 (69.1 %) were negative. On multivariate analysis, metformin treatment was associated with a significant reduction in CDI [odds ratio (OR)=0.58; 95 % confidence interval (CI), 0.37–0.93; p=0.023], while heart failure was associated with significantly higher rates of CDI (OR=1.654; 95 % CI, 1.007–2.716; p=0.047), together
N. Eliakim-Raz (*) : D. Yahav : E. Goldberg : H. B. Zvi : J. Bishara Infectious Diseases Unit, Rabin Medical Center, Beilinson Hospital, Petah Tikva, Israel e-mail:
[email protected] G. Fishman : G. Y. Stein Department of Internal Medicine B, Rabin Medical Center, Beilinson Hospital, Petah Tikva, Israel N. Eliakim-Raz : D. Yahav : E. Goldberg : G. Y. Stein : A. Barsheshet : J. Bishara Sackler Faculty of Medicine, Tel Aviv University, Ramat Aviv, Israel A. Barsheshet Department of Cardiology, Rabin Medical Center, Beilinson Hospital, Petah Tikva, Israel
with poor functional status, previous hospitalization, and abdominal surgery. Our findings suggest that, in diabetic patients, in addition to the well-recognized risk factors, heart failure is an additional risk factor for CDI, while metformin treatment seems to have a protective effect against the development of CDI. The exact mechanisms underlying this protective effect remain to be fully understood.
Introduction Clostridium difficile is a Gram-positive, spore-forming anaerobic bacillus and a member of the Clostridium genus (part of the Firmicutes phylum). It is the most common cause of infectious nosocomial diarrhea among adults in developed countries [1–3]. In recent years, this organism has demonstrated increasing virulence with the emergence of new strains, resulting in increased morbidity and mortality [4–6]. C. difficile infection (CDI) is associated with increased length of hospital stay and healthcare costs [7, 8]. Many risk factors have been described for developing CDI, including the use of antimicrobial agents, advanced age, hospitalization, severe comorbidities, exposure to cytotoxic chemotherapy, immunosuppressive therapy, and use of proton pump inhibitors [9–11]. CDI is believed to be causally related to perturbations in the intestinal microbiota [12], with an increased Firmicutes/ Bacteroidetes ratio in the intestinal microbiota. Recent evidence showed that type 2 diabetes mellitus is an important risk factor for the recurrence of CDI [13]. Chang et al. demonstrated that patients with recurrent CDI had decreased species of Bacteroidetes and Firmicutes in their stool compared to patients with just a single episode of CDI [14]. On the other hand, recent studies have suggested a changed
Eur J Clin Microbiol Infect Dis
ratio of Bacteroidetes and Firmicutes in diabetes gut microbiota [15, 16]. According to Larsen et al.’s study, the ratio of Bacteroidetes to Firmicutes increases in diabetics [15]. For the past 50 years, metformin, an oral antidiabetic biguanide, has been widely used to treat type 2 diabetes, particularly in overweight and obese patients with normal kidney function [17]. Metformin decreases hyperglycemia primarily by suppressing hepatic gluconeogenesis, the molecular mechanism of which is still incompletely understood. Recent human and animal studies have shown that gut microbiota and their metabolic pathways are influenced by metformin treatment, causing increased levels of Enterobacteriaceae and decreased levels of Clostridium and Eubacterium [18], and an increase in the abundance of Akkermansia muciniphila [19]. To the best of our knowledge, predicting factors for CDI in diabetic patients and the effect of metformin therapy on CDI have never been assessed. We hypothesized that metformin treatment may provide protection against CDI in diabetic patients. For this purpose, we conducted a retrospective study to assess risk factors for CDI in diabetic patients and determine whether metformin therapy protects against the development of CDI in these patients.
Methods Subjects and study design This is a retrospective, case–control study. We identified, from the microbiology laboratory records from January 2009 to December 2013, all consecutive diabetic patients hospitalized in our center who were tested for C. difficile toxins. The cohort was split into two groups: diabetic patients with a positive test for C. difficile (CD-positive group) and diabetic patients with a negative test for C. difficile (CD-negative group). Patients were included only once in the analysis. In our hospital, the C. difficile toxin assay can be performed on unformed stool specimens only. All patients had diarrhea and had been taking antibiotic therapy during the previous 90 days. We excluded patients who were ≤18 years old, as well as those who lacked sufficient data for calculating the Charlson comorbidity index score [20]. Data on demographic characteristics, underlying conditions, previous hospitalizations during the previous 6 months, abdominal and other surgery, and prior medications use including metformin were collected for all patients, and characteristics of CDI at presentation, therapy for CDI, and clinical and laboratory findings were collected for all patients in the CD-positive group. All data were obtained from the patients’ electronic medical records. The control group included consecutive patients with diabetes mellitus, a negative stool test for C. difficile toxins, and who were hospitalized during the same period and in the same departments. The study was approved by the hospital’s ethics
committee. Informed consent was not required, given the retrospective and non-interventional nature of the study. Definitions CDI was defined as diarrhea not attributed to any other cause and associated with a positive stool test for C. difficile toxins. The decision to test for C. difficile was made by physicians uninvolved in the study. A positive test was defined by the C. DIFF QUIK CHEK COMPLETE assay (TechLab, Blacksburg, VA), which simultaneously detects both glutamate dehydrogenase (GDH) antigen and toxins A and B. Discrepant results between GDH and toxin A/B were solved using the Xpert C. difficile polymerase chain reaction (PCR) assay (Cepheid, Sunnyvale, CA), which is a multiplex realtime PCR that detects the toxin B gene (tcdB), the binary toxin gene (cdt), and the tcdC gene deletion at nt 117 [9]. Discrepant but PCR-positive patients were considered positive. Statistical methods Continuous variables are expressed as mean ± standard deviation (SD). Categorical data are summarized as frequencies and percentages. Characteristics of patients were compared between the groups, stratified by CDI, using the t-test for continuous variables and the Chi-square test or Fisher’s exact test for dichotomous variables, as appropriate. The association of baseline metformin treatment and CDI during hospitalization was investigated using multivariate logistic regression models. Covariates included in the multivariate model were identified using a best subset procedure among 20 relevant clinical variables. The eight variables selected in the multivariate model included metformin treatment, heart failure, age, Charlson comorbidity index score, pressure ulcers, previous hospitalization, previous abdominal surgery, and bedridden patients. All p-values were two-sided, and a p-value
