Surveillance and outbreak report Hospital-based Clostridium difficile infection surveillance reveals high proportions of PCR ribotypes 027 and 176 in different areas of Poland , 2011 to 2013

H Pituch 1 , P Obuch-Woszczatyński 1 , D Lachowicz 1 , D Wultańska 1 , P Karpiński 1 , G Młynarczyk 1 , SM van Dorp 2 , EJ Kuijper 2 , the Polish Clostridium difficile Study Group 3 1. Department of Medical Microbiology, Medical University of Warsaw, Warsaw, Poland 2. Department of Medical Microbiology, Leiden University Medical Centre, Leiden, the Netherlands 3. The members of the group are listed at the end of the article


Introduction
Clostridium difficile infection (CDI) is a common nosocomial problem, which can affect patients following antibiotic treatment [1].Since 2003, reports of outbreaks of severe CDI have increased in Canada and the United States [2][3][4].This increase coincides with the emergence and rapid spread of a more virulent strain of C. difficile belonging to the North American Pulsotype 1/BI, which is referred to in Europe as polymerase chain reaction (PCR)-ribotype 027 [5].Some of the characteristics of this strain are higher in vitro production of toxins A (TcdA) and B (TcdB) and the presence of a third toxin called the binary toxin.The increase in toxin production is related to two mutations in the toxin regulatory gene tcdC: an 18 base-pair (bp) deletion, and deletion at position 117 [6].In Europe, the epidemic strain was first observed in Belgium, France, the Netherlands, and the United Kingdom [7][8][9][10] and most recently caused outbreaks in Austria, Portugal and Romania [11][12][13][14].Outbreaks of CDI caused by PCRribotype 027 have been associated with fluoroquinolone use in particular, and circulating PCR-ribotype 027 clones exhibit high levels of resistance against newer-generation fluoroquinolones [15].The first Polish isolate of C. difficile PCR-ribotype 027 was detected in 2005 and a closely related PCR-ribotype 176 was discovered in 2008 [16,17].CDI outbreaks associated with ribotypes 027 and 176 have been documented in three hospitals in Poland between 2008 and 2010 [18].
Based on participation in the European Clostridium difficile Infection surveillance Network (ECDIS-Net), which is a European Centre for Disease Prevention and Control (ECDC)-supported programme to build capacity for CDI surveillance in Europe, we previously set up a surveillance network of hospital-based laboratories in Poland.Here, a new network made up of a number of randomly selected Polish hospital-laboratories was constructed to conduct surveillance from 2011 to 2013.The aim of the study was to determine the annual CDI incidence rates in these institutions.In addition, periodical microbiological surveillance (February-March in both 2012, and 2013) was conducted to characterise C. difficile isolates obtained in the same hospitals.

Selection of hospitals
The aim was to include secondary and tertiary care hospitals.Funding only allowed to include a maximum of 20 hospitals in Poland, so an invitation to participate in the study was sent to 20 clinical hospital-laboratories which were selected at random among 600 healthcare facilities in different parts of the country.The number of people living in the areas of the 20 hospitals-laboratories is 10,867,100.Of the 20 hospital-laboratories contacted, seven declined to participate.Reasons for not enrolling in the study included not performing CDI surveillance (n=2 laboratories) or insufficient capacity (n=1 laboratory).In some cases (n=4), the reasons were not listed.Of the 13 hospitals (designated H1 to H13) that responded favourably, 11 provided secondary (n=5) or tertiary care (n=6), and two were specialised in pulmonology/thoracic surgery (H1) and oncology (H8).The number of beds among the hospitals varied from 250 (H1) to 1,310 (H13).Although the hospitals did not cover all Polish provinces, and three were in Warsaw, the 13 hospitals were located in 10 different cities across Poland, namely Bystra (H1), Bydgoszcz (H9), Krakow (H13), Łańcut (H3), Maków Mazowiecki (H2), Piła (H7), Płock (H10), Poznań (H4), Szczecin (H5), Warsaw (H8, H11, and H12), and Włocławek (H6).

Data collection
Before the start of the study (in January 2012), surveys were sent to participating hospital-laboratories, with requests for epidemiological data in order to calculate the annual CDI incidence rates for 2011, 2012 and 2013.Questions about C. difficile diagnostic testing were also asked.The surveys were completed by early 2014.In addition, between 1 February and 31 March in the two consecutive years 2012 and 2013 respectively, participating hospitals sent strains from patients identified with CDI to the Anaerobe Laboratory in Warsaw for molecular characterisation.

Determining incidence rates of Clostridium difficile infections
The study design was a hospital-based surveillance, using CDI case definitions based on ECDIS-Net recommendations as previously described by Kuijper et al. [19].Hospitalised patients were included as a CDI case if onset of symptoms (abdominal pain, diarrhoea, ileus, toxic megacolon) occurred within the surveillance period.The detection of patients with CDI was based on the finding of clinical specimens testing positive for C. difficile in the laboratory.Annual hospital incidence rates were calculated per 10,000 patientdays.Numerator data included all reported initial CDI episodes of hospitalised patients above the age of two years, as well as recurrent episodes that occurred more than eight weeks after the onset of a preceding

Diagnostic tests used for Clostridium difficile infection and indications for testing
The epidemiological surveys also comprised questions on C. difficile laboratory diagnostic testing, and indications for testing.Participating laboratories were asked to report the type of screening test such as enzyme immunoassay for TcdA only, TcdA and/or B or glutamate dehydrogenase (GDH), molecular tests, toxigenic culture, or any other tests.Subsequently, participants were asked if they used a confirmation test.For both questions, there was a possibility to report more than one test.
Furthermore, decision criteria to perform C. difficile diagnostic testing were assessed, i.e. testing based on a physicians' request, testing in cases of antibioticassociated diarrhoea, testing all diarrhoeal stools, or testing of diarrhoeal stools in a hospitalised patient from the third day of admission (nosocomial diarrhoea).

Molecular characterisation of isolates
Faecal specimens sent by the clinicians for routine C. difficile detection were tested in hospital-laboratories according to their standard methodology.All C. difficile strains (max 30) isolated from consecutive faeces samples testing positive for CDI in February and March of 2012 and 2013, respectively, were sent to the Anaerobe Laboratory, Medical University of Warsaw for detection of toxin encoding genes and PCR-ribotyping.Only one sample per patient was included in the study.Faecal samples were inoculated anaerobically on selective media for 48 h or 24 h, and C. difficile colonies were sub-cultured on blood agar and identified using standard methods, as described previously [18].
The toxigenicity was characterised by testing C. difficile isolates for tcdB and binary toxin encoding genes using the GeneXpert CD assay (Cepheid; Sunnyvale, California, United States), which is based on a real-time PCR method.PCR ribotyping was performed according to the method described by Stubbs et al. [20].The Cardiff-ECDC collection of reference isolates (n = 23) of C. difficile was used as a reference set.

Clostridium difficile infection incidence
During the three year-surveillance period, the annual mean incidence for the collaborating hospitals was 8.17 CDI per 10,000 patient-beds.1).
The highest incidence rates of CDI were observed in university hospitals, for example, H4 (range: 4.7-5.2 per 10,000 patients-days), H12 (range: 12.4-19.3per 10,000 patients-days), and H13 (range: 5.2-7.1 per 10,000 patients-days), and the lowest in provincial hospitals such as H6 (range: 0.2-3.8 per 10,000 patients-days) and H7 (range: 0.9-4.7 per 10,000 patients-days).Different decision criteria were applied to perform diagnostic tests for CDI on faeces specimens.Two of the total 13 laboratories tested all diarrhoeal faecal samples submitted to the laboratory.Six tested specimens only on the request of a physician and five applied additional criteria for CDI diagnostics, such as testing samples in case of antibiotic-associated diarrhoea and testing all diarrhoeal samples from patients who developed diarrhoea more than two days after admission (nosocomial diarrhoea).
A high percentage (≥75%) was found in several hospitals located in different cities H3, H4, H9, H10 and H13.However, differences were observed between hospitals in the same city.The distributions of C. difficile PCRribotypes in 2012 and 2013 are shown in the Figure.

Discussion
Stimulated by the ECDC capacity-building network for CDI surveillance (ECDIS-Net) we developed a surveillance programme to estimate the incidence of CDI in hospitalised patients in Poland, comprising annual epidemiological surveys and periodical molecular surveillance.The main objective of the Polish surveillance was to encourage local laboratories to develop local diagnostic algorithms and to support surveillance studies that use internationally agreed-upon definitions.We found an annual mean incidence of 8.17 CDI per 10,000 patient-beds during the three year-surveillance period in the collaborating hospitals.The CDI incidence rate seems to have increased from  In our study, C. difficile PCR-ribotype 027 was prevalent in all participating hospitals.A particularly high incidence was observed in university hospitals, H4, H11, H12, and H13.The CDI incidence varied considerably among the participating hospitals, not only related to the hospital sample size, but also due to the background of the hospitals, such as university or provincial hospitals with specific services (e.g.transplant medicine, haematology) or specialised hospitals (pulmonology/thoracic surgery and oncology).
The high incidence of PCR-ribotype 027 strains in these hospitals is likely a reflection of multiple exposures to the environment of healthcare facilities, antibiotic consumption and disruption of intestinal microbiota, and immunosuppression.However, we did not analyse antibiotic consumption among patients in this study.We observed that Polish university hospitals experienced higher number of CDI episodes compared with provincial hospitals.
After  [23].We also detected C. difficile PCR-ribotype 018, which is the most frequently found ribotype in Italy [27].We found seven non-toxigenic isolates belonging to six uncommon PCR ribotypes.It is likely that these isolates were derived from patients with mixed infections of both toxigenic and non-toxigenic isolates.Other PCR-ribotypes were detected sporadically, i.e., once or twice, during the two study periods.
Our study has a few limitations.First, of the 20 hospital-based laboratories invited, only 13 laboratories participated.This may be attributed to the voluntary nature of participation of the survey and lack of funding, but may have resulted in selection bias.Second, our study also included three smaller hospitals with 260 to 265 beds, which influenced the precision of our calculated incidence rates.Overall, the results of this surveillance programme were not yet validated.Lastly, we could only characterise a part of the C. difficile strains from patients with diagnosed CDI in the participating hospitals.An important achievement of our study is the construction of a network to survey CDI in Poland.Hospitals collect a minimum amount of clinical and epidemiological data and send their isolates to a central laboratory.Our next steps are to validate the surveillance programme, to standardise the diagnostics of CDI and optimise patient selection for CDI testing.The identification of the (re)emergence of PCR-ribotypes 027 and PCR-ribotype 176 through molecular surveillance in this study is of concern and needs to be addressed through a national approach to combat CDI.Further studies evaluating the virulence factors and epidemiology of PCR-ribotypes 027 and 176 are urgently needed.Our study underscores the need for local and regional surveillance in Poland to detect and control CDI.

TABLE 1
List of collaborating hospital-laboratories included in the Polish surveillance programme for Clostridium difficile infection and annual incidence rates, Poland, 2011-2013 episode.Age and sex of patients with CDI were registered.Denominator data comprised reported annual numbers of admissions and patient-days per hospital (in 2011, 2012, and 2013).The incidence rates of all participating hospitals were used to calculate a mean incidence rate.

Table 2
).The C. Diff Quik Chek Complete is one test but recognises two different targets and can therefore be considered as a two-step algorithm.Three laboratories used only an enzyme immunoassay for Tcd A/B detection.In addition, one laboratory used the Illumigene C. difficile Kit (Illumigene C. difficile CDI: Clostridium difficile infection; EIA: enzyme immunoassay; GDH: glutamate dehydrogenase; TOX A/B: toxins A and B. Laboratory tests were named as follow: EIA GDH (TechLab, USA): EIA to detect GDH; EIA GDH and EIA TOX A/B: EIA test for GDH alone and EIA confirmation test for TOX A/B; EIA TOX A/B: different EIA to detect toxins A and B (mainly TOX A/B, Wampole, USA); GDH+TOXA/B: combined test detecting both TOX A and/ or B and GDH (The C. Diff Quik Chek Complete (TechLab; Blacksburg, VA, USA and Alere; Waltham, MA, USA)); TC: toxigenic culture; qPCR: The Xpert kit (Cepheid, Sunnyvale, CA, USA). a Hospitals where two different tests were used for screening and confirmation.b Indicates whether diagnostic changes occurred per hospital in 2012-2013 (TC, GDH+TOXA/B; qPCR).

Clostridium difficile isolates
A total of 13 hospital-laboratories (one laboratory per one hospital; 11 in 2012 and an additional two new laboratories in 2013) participated in the two-month periods of molecular surveillance and sent a total of 166 C. difficile isolates to the central laboratory.Of

TABLE 3
Proportion of PCR-ribotype 027 per toxigenic strains in hospital-laboratories participating in the surveillance programme for Clostridium difficile infection, Poland, 2012 and 2013 (n=159 strains)