Infections caused by Clostridium difficile (C. difficile) (CDI) are generally diagnosed by the presence of persistent liquid diarrhea, with the identification of C. difficile in a stool sample using some of the different tests available. A diagnosis based on colonoscopic or histopathological findings evidencing the presence of pseudomembranous colitis is less common.1 The importance of such infections is reflected by the fact that C. difficile may be present in 20–30% of the cases of diarrhea associated to antimicrobial use.2 Few epidemiological data are available on the prevalence of CDI in hospitalized patients, and there is even less information referred to critical patients admitted to Departments of Intensive Care Medicine or Intensive Care Units (ICUs). A study made between 1997 and 2005 in Canadian hospitals recorded an incidence of between 3.4 and 8.4 cases per 1000 admissions in acute care hospitals, and of 3.8–9.5 cases per 10,000 patients-day.3,4 The mortality rate attributed to CDI is low (less than 2%),4,5 though the associated excess costs in American hospitals during the period 2000–2002 was estimated to be 3200million USD/year.6

The risk factors underlying CDI include old age (>64 years),7,8 the duration of hospital stay,9 exposure to antimicrobial drugs,10 cancer chemotherapy,11,12 immune depression related to HIV infection,13 gastrointestinal surgery,14 enteral diet administered through a nasogastric tube,15 and the administration of acid-suppressing drugs (antihistamines and proton pump inhibitors).16 Many of these factors are present in critical patients admitted to the ICU. However, infection registries in such Units tend to focus on infections related to the use of invasive devices17–the information on CDI being scarce and limited to the description of epidemic outbreaks and mortality or cost increment estimates.18–23 The objective of this study is to determine the characteristics of critical patients with CDI admitted to Spanish ICUs in the year 2012, independently of the origin of the infection, and to describe the factors related to the development of CDI and the treatment strategies employed.

The study included 68 patients with CDI in 33 (19.1%) ICUs of the 173 Units were participating in the ENVIN-ICU registry (2.1 patients per ICU with CDI; range 1–6). The distribution of cases over the months is shown in Fig. 1. The CDI episode rates referred to the national participation period in the registry (April–June) were 1.59 per 1000 patients admitted to the ICU, and 2.06 episodes per 10,000 days of stay in the ICU (patients-day). Forty-five patients were males (66.2%), with a mean age of 63.4 (16.4) years, an APACHE II score upon admission of 19.9 (7.4), and a predominantly clinical background disease condition in 44 cases (64.7%) (Table 1).

Figure 1.

Distribution of the cases of Clostridium difficile infection (CDI) detected in the Intensive Care Units in the course of the months of the year (31 patients in the national vigilance period).

(0.06MB).

In 62 cases (91.2%), CDI was associated to clinical manifestations in the form of liquid depositions, and in 40 cases (58.8%) a systemic inflammatory response was present in the form of severe sepsis or septic shock. In one case CDI evolved toward toxic megacolon requiring emergency surgery for control and treatment. In 13 cases (19.1%), CDI was classified as being community acquired, while another 13 (19.1%) were of nosocomial out-ICU origin, and 42 (61.8%) were nosocomial in-ICU infections (Table 2).

Table 3 shows the comorbidities and invasive devices in the patients with CDI. Of note is the great need for monitoring techniques (Table 4). The specific factors related to CDI were a patient age of >64 years in 39 subjects (57.4%), admission to hospital in the preceding three months in 31 (45.6%), the use of antimicrobials in the 7 days before the diagnosis of CDI in 57 (83.8%), the use of enteral nutrition in 23 (33.8%), and the prescription of H2 inhibitors in 39 patients (57.4%). In contrast, chemotherapy (7 patients, 10.3%) and gastroduodenal surgery (5 patients, 7.4%) were much less frequent. Forty-four patients with CDI (64.8%) presented three or more specific risk factors for this infection. Among the groups of antimicrobials used before the diagnosis of CDI, mention must be made of betalactams associated to betalactamase inhibitors (36.8%), carbapenems (33.8%) and cephalosporin (23.5%).

Specific treatment for CDI was indicated in 66 patients (97.1%) (Table 5) in the form of a single antibiotic in 42 cases (61.8%) and two or more in 24 (35.3%)–though in only 18 cases were both drugs administered simultaneously. Metronidazole predominated, with 60 (88.2%) cases, followed by vancomycin in 31 (45.6%). The use of these antibiotics, specifically prescribed for the treatment of CDI, was accompanied by the use of other antimicrobials for the treatment of other infections in 24 cases (35.3%). Seven patients (10.3%) presented adverse effects probably or possibly related to the administration of antimicrobials specific for CDI, including hematological disorders (2 cases) and water–electrolyte alterations (2 cases) in relation to the administration of metronidazole, and a case of severely altered liver function associated to the simultaneous use of metronidazole and vancomycin. In one case the initial treatment was changed because of the adverse effects.

The clinical response was considered satisfactory (cure and/or improvement) in 51 cases (75.0%), while no changes or failures were recorded in 10 (14.7%). The response was not evaluable in another four cases (5.9%). Negative microbiological conversion was achieved in 37 cases (61.7%), follow-up was not conducted in 20 (29.4%), and persistence was detected in two subjects (2.9%). In the patients treated with metronidazole in monotherapy (n=36), the satisfactory response rate was 77.8%, versus 100% with vancomycin (n=7), and 61.1% with both antibiotics (n=18). Only two cases of reinfection were recorded. The overall in-ICU mortality rate was 25.0% (n=17), while another two patients died in hospital after discharge from the ICU (overall in-hospital mortality rate 27.9%; n=19). In no case could mortality be related to CDI (Tables 5 and 6).

The main contribution of this study has been the description of the critical patients admitted to Spanish ICUs with Clostridium difficile infection (CDI). These patients are generally elderly, with background clinical disease and multiple comorbidities, a high degree of severity upon admission to the ICU, the need for broad-spectrum antibiotic treatment, and with abundant liquid depositions in the first week of stay in the ICU. In 62% of the cases the ICU was found to be the origin of CDI.

The estimated CDI rates in the months of April to June were 1.6 cases per 1000 patients admitted to the ICU and 2.1 episodes per 10,000 stays in ICU (patients-day). Most studies in patients admitted to the ICU refer to epidemic outbreaks or endemic situations in one or more centers, or conduct an analysis of the impact of CDI upon stay, costs and mortality in an ICU characterized by a high proportion of patients with this infection. Recently, Bouza et al.,18 on occasion of the European Congress of Clinical Microbiology and Infectious Diseases of the European Society of Clinical Microbiology and Infectious Diseases (April 2013, Berlin, Germany), reported the CDI rate corresponding to the last 10 years in the four ICUs of a general hospital. The global incidence of CDI was 13.2 cases per 10,000 days of stay in the ICU, and the figure showed a clear tendency to decrease in recent years (from 15.9 in 2003 to 8.8 in 2012). Likewise, the authors demonstrated the existence of important differences in rates depending on the characteristics of the ICU–the highest rates being found in clinical ICUs attending patients following heart surgery. Zahar et al.19 in turn published a retrospective study on the clinical impact of CDI in patients admitted to three ICUs. The rates recorded in their study were 13.1 cases per 1000 patients (calculated figure) and 3.6 cases of CDI acquired in the ICU per 1000 patients-day. Lawrence et al.22 conducted a retrospective analysis of the impact of C. difficile colonization as a risk factor for the development of CDI. They documented a rate of 35.3 new cases per 1000 patients admitted for over 24h to the ICU (calculated figure). In both of the aforementioned studies the rates were higher than those recorded in our registry. This may be because we included ICUs with very distinct characteristics and risks, and without endemic conditions for this particular pathogen. In fact, only one out of every 5 ICUs (19%) that contributed data to the registry identified cases of CDI.

A possible reason for the reduced diagnosis of CDI in our setting may be the lack of a pre-established study method. A study conducted in a Spanish hospital has demonstrated that the application of a control and vigilance program doubles the recorded CDI rate.25 In this sense, homogenization of the study methodology in cases of suspected CDI has been advocated at European and American level, proposing a search for the infection in all patients admitted to the ICU with liquid diarrhea, defined as three of more liquid depositions in the course of 24 consecutive hours (or less).26,27 The watery, liquid or unformed stool samples should be quickly sent to the Microbiology laboratory for routine study, including the investigation of C. difficile.

In our study, the tests used for the diagnosis of CDI have been those routinely employed in each Microbiology laboratory, with a predominance of stool cultures, the determination glutamate dehydrogenase, enzymoimmunoassay for toxins A and B, and, in fewer cases, polymerase chain reaction (PCR). In order to homogenize the identification of CDI, a diagnostic protocol (2-step algorithm) has been proposed, with due adaptation to the possibilities of each individual laboratory.26,27 A first determination is made of glutamate dehydrogenase, followed in positive cases by enzymoimmunoassay for toxins A and B. Only when these two techniques prove negative is a more specific third technique suggested (PCR or toxigenic cultures). As a less expensive alternative, this step can be replaced by stool culture, followed in positive cases by direct toxin determination in the isolates.

The treatment used mainly consisted of monotherapy with metronidazole and, in fewer cases, with vancomycin. The preferential choice of metronidazole evidences that the form of presentation of CDI is regarded by the clinicians as mild or moderate, since in the more severe or complicated presentations oral (or rectal, in the case of paralytic ileus) vancomycin is recommended, with or without metronidazole simultaneously.1 A recent review28 has analyzed the studies that established comparative assessments of the efficacy of the different antimicrobials used to treat CDI. All of them were of low to moderate statistical power, and no differences were observed among the treatments. It should be mentioned that up to one-quarter of all the therapies used in our series were administered as the combination of both antibiotics, despite a lack of evidence warranting their effectiveness. In 7 of the cases (10%), adverse reactions attributed to the antibiotics used to treat CDI were recorded.

The clinical response to the treatment provided was rated as satisfactory (cure or improvement) in 75% of the cases. In the rest of the cases evaluation was not possible or no changes were observed in patients who died as a result of progression of their background disease condition. The overall in-ICU mortality rate was 25%, and in no cases could death be related to CDI. Different studies have examined the impact of the presence of CDI upon mortality among critical patients. Most of them show the presence of CDI to be associated to high overall in-ICU and in-hospital mortality19–22,29; however, on conducting comparative analyses (involving patients with diarrhea without CDI or case-controls), no relationship was observed between the presence of CDI and patient mortality.19,29

The reported frequency of CDI recurrence is between 6 and 25%.1 The recurrence rate in our study was low. This may be because a quarter of the patients have died, and follow-up was only made until discharge from hospital.

This study has important limitations, such as its retrospective nature based on the analysis of reported cases in the context of a voluntary national registry. The registry of CDI was first included in the year 2012, and this may have caused omissions in recording the infection. On the other hand, no common protocol was used to request microbiological studies in patients at risk, and the tests applied to the stool samples varied from one hospital to another. Likewise, we have not been able to establish a relationship between the presence of CDI and outcome.

In sum, the present study contributes information on the characteristics of the patients with CDI admitted to Spanish ICUs, and on the treatments provided. Of note in this respect is the fact that two antibiotics were administered simultaneously in one-quarter of the cases. The calculated infection rates were lower than those published by other authors and fell short of the rates expected from the CDI-specific risk factors present in these patients–thus evidencing that C. difficile infection is underdiagnosed in critically ill patients admitted to the ICU. A consensus-based study methodology is needed (collection and submission of samples and specific tests in the Microbiology laboratories) in reference to patients with liquid depositions in our setting.

M. Palomar has received an international grant from Pfizer for maintenance of the ENVIN-ICU registry. F. Alvarez-Lerma has received a grant from Astella-Farma for the collection and analysis of the data referred to Clostridium difficile infection.

The authors declare that they have no conflicts of interest.