Patient population, CDI case definition, and analysis of risk-factors
The current study was initiated after recognition of a possible CDI outbreak in the first quarter 2015 on the septic ward of the department of orthopedic surgery. This department with three independent wards is part of an 1,100 bed tertiary care university hospital and acts as a regional referral center for the treatment of osteoarticular infections. Patients with osteoarticular infections stay on a specialized ward to which we refer as “septic ward”.
For the study hospital-acquired CDI was defined as new onset of diarrhea 48 h after hospital admission and laboratory-confirmed detection of C. difficile toxin genes by PCR. Severe CDI was defined by the presence of at least one of the following symptoms: fever >38.5 °C, decreased kidney function (creatinine >1.5x103g/L) and/or high leukocyte count (>15 × 109 cells/L). A recurrent CDI case was defined as new onset of CDI within 8 weeks after resolution of a previous CDI episode. No children were included in the study. An outbreak situation was defined as two or more CDI cases on a single ward within 28 days. Risk factors for infection with C. difficile RT027 were evaluated by comparing patients with CDI caused by RT027 with patients infected with non-RT027 ribotypes.
Infection control and antibiotic stewardship
Infection control measures were implemented after recognition of an enhanced CDI incedence. The infection control bundle included (i) handwashing with soap after disinfection, to reduce first vegetative forms of C. difficile or other bacteria and secondly spores of C. difficile, (ii) use of hygienic bags for bedpans, (iii) strict isolation or cohorting of infected patients, (iv) personalized use of all materials with direct contact to the patient e.g. stethoscope, and (v) disinfection of surfaces and equipment in all wards, operating rooms and physiotherapy with peracetic acid and peroxide hydrogen vaporization after discharge of patients. Intensified environmental cleaning procedures were controlled by sampling of surfaces as described below. Additionally, health care workers were educated on diagnosis, treatment and prevention of CDI and provided with an antibiotic risk checklist which groups of antibiotics in 3 risk categories (low risk: linezolid, vancomycin, metronidazole, tetracycline, trimethoprim/sulfamethoxazole, fosfomycin; daptomycin, medium risk: all beta-lactam-antibiotics with the exception of third generation cephalosporins; high risk; clindamycin, third generation cephalosporins, fluoroquinolones). Physicians were instructed to follow the in-house guidelines for antibiotic therapy and microbiological diagnostic of osteoarticular infections and to avoid the prescription of high risk antibiotics for therapy. In quarter 3/2015 an intensified antibiotic stewardship program was implemented which included a weekly review of all antibiotic prescriptions by an orthopedic surgeon and a clinical microbiologist trained in antibiotic stewardship. The organization of the team was not changed since the beginning of the intervention. If possible antibiotic therapy was adjusted in order to avoid CDI high risk antibiotics as described above.
In order to control the impact of infection control measures and antibiotic stewardship the CDI incidence was monitored as number of cases per 1,000 hospital bed days per quarter.
In addition the impact of the antibiotic stewardship intervention was controlled by quarterly monitoring of antibiotic consumption on the septic ward. Antibiotic consumption was calculated as defined daily doses (DDD) per 100 hospital bed days according to WHO standards .
Diagnostic specimens and CDI laboratory diagnostic testing
Diagnostic stool specimens were collected from 63 adult patients showing symptoms of diarrhea between June 2014 and December 2015. Stool diagnostic was performed as part of the routine microbiological diagnostic in a two-step algorithm.
First, clinical samples were primarily screened with a Clostridium glutamate dehydrogenase (GDH)-specific enzyme-linked immunosorbent assay (RIDASCREEN® Clostridium difficile GDH, R-Biopharm, Darmstadt, Germany) using the protocol provided by the manufacturer. In case of positive results (n = 29) DNA was extracted from the original stool sample. In brief, 100 μl stool in 900 μl sample buffer were used, followed by centrifugation at 1,000 x g for 5 min. Subsequently 400 μl supernatant was transferred into Precellys® Soil grinding SK38 tubes (Bertin Technologies, USA) and homogenized by centrifugation (5,000 x g for 75 sec) using the MagNA Lyser System (Roche Diagnostics, Mannheim, Germany). The lysate was clarified by centrifugation at 1,000 x g for 5 min and subsequent incubated for 10 min at 70 °C in a thermoshaker. DNA-extraction was performed using the QIAamp® DNA Mini Kit (Qiagen, Hilden, Germany) according to manufacturer’s instructions.
Second, purified DNA samples were tested for the presence of C. difficile DNA using two commercially available real-time PCR test systems. C. difficile toxins A (tcdA) and B (tcdB) genes were detected using the RealStar® Clostridium difficile PCR Kit 1.0 (altona Diagnostics, Hamburg, Germany). The binary toxin gene and the Δ117 deletion in the tcdC gene were detected using Xpert® C. difficile/Epi PCR assay (GeneXpert, Cepheid, Sunnyvale, CA, USA). All tests were performed according to manufacturer’s protocol.
Multilocus sequence typing (MLST)
MLST was performed with DNA either isolated from feces (n = 10) or from isolated strains (n = 16). The MLST was performed as described before , targeting 7 housekeeping genes of C. difficile: adk, atpA, dxr, glyA, recA, sod and tpi. The sequencing reactions were run on a 3130xl Genetic Analyzer (Applied Biosystems). Editing, alignment, and phylogenetic analysis of sequences were performed with the program MEGA 6.0 . DNA sequences were uploaded to the MLST database and C. difficile sequence types (ST) were received from the website .
Clostridium difficile culture and susceptibility testing
Clostridium difficile was cultured form GDH-positive stools (n = 16) using a chromogenic medium (chromID™ C. difficile, bioMérieux, Marcy l’Etoile, France). The medium was inoculated with 10 μl feces and incubated anaerobically at 35 ± 1 °C for 7 days. Presumptive C. difficile colonies were confirmed by MALDI-TOF MS (VITEK® MS, bioMérieux).
The minimal inhibitory concentrations (MIC) of metronidazole, vancomycin, rifampicin, levofloxacin, and clindamycin were determined by gradient strip test (Etest, bioMérieux) on Brucella blood agar (Becton Dickinson, Heidelberg, Germany) inoculated with 100 μl solution of a 1.0 McFarland suspension of C. difficile in saline as described previously . Agar plates were incubated under anaerobic conditions at 35 ± 1 °C for 48 h. For the interpretation of MIC the EUCAST epidemiological cut off values (ECOFF) were used for metronidazole (>2 mg/L), vancomycin (>2 mg/L), rifampicin (>0.004 mg/L). No ECOFF are available for clindamycin and levofloxacin.
Enviromental sampling of inpatient environment
Sampling of the inpatient environment for the presence of C. difficile was performed as described . Briefly, surface samples were taken using 25 cm2 sponge swabs pre-moistened with neutralizing solution (Lab M Ltd, Heywood, United Kingdom). Frequent contact surfaces of the patient room (head/foot-end boards of patient beds, bed rail, bedside table, nurse call button, patients telephone) and the en-suite bathroom (toilet seat, toilet assist handle) were sampled. After sampling sponge swabs were placed aseptically into the sterile sample transport bag prefilled with 10 ml neutralizing solution. The laboratory bags were opened and supplemented with 40 ml sterile phosphate-buffered saline (Becton Dickinson, Heidelberg, Germany) to yield a final volume of 50 ml. Sponge bags were resealed and homogenized manually by massaging the bag for 1 min. After 10 min. incubation at room temperature the whole volume was passed through a 45 μm membrane filter (Pall GmbH Laboratory, Dreieich, Germany). Filters were aseptically put onto Brazier's CCEY agar (Oxoid, Wesel, Germany) and incubated 48 h in an anaerobic atmosphere at 35 ± 1 °C. After 48 h, suspected C. difficile colonies were further analysed by MALDI-TOF MS. Confirmed C. difficile isolates were tested for toxin production and typed by MLST.
Ribotyping and multiple-locus variable-number tandem repeat analysis (MLVA)
Ribotyping and MLVA analysis of 11 ST1 isolates was performed by the National Consultant Laboratory for C. difficile in Homburg/Saar (University of Saarland Medical Center, Homburg, Germany). PCR-ribotyping was performed for each isolate according the standard protocol (European harmonized diagnostic procedures ECDIS; http://www.ecdisnet.eu) as described earlier  which included capillary gel electrophoresis of fluorescent labelled fragments (Beckmann Coulter, Brea, California USA) and ribotype assignment to an institutional databank by an automated software tool (BioNumerics version 7.1, Applied Math, Sint-Martens-Latem, Belgium). MLVA was carried out as described previously  with BioNumerics as automated software (version 7.1, Applied Math, Sint-Martens-Latem, Belgium). The definition of clonality was based upon previous studies with a genetic difference of less than 3 repeats while a clonal cluster was defined by ≤2 repeat differences and genetic related isolates by ≤10 repeat differences .
Statistical analyses were performed using the Fisher’s exact Test for bivariate analysis of C. difficile risk factors (non RT027 CDI versus RT027 associated CDI) and the two-sample independent t Test for the means of antibiotic consumption data before and after intervention with a significance level of p values <0.05. The Microsoft Excel statistic tool, OpenEpi (http://www.openepi.com) was used to analyze the data. The upper and lower statistical boundaries were defined as the mean CDI incidence of the four previous quarters plus/minus standard deviation (MA ± SD).