Clinical Significance of Clostridium difficile in Children Less Than 2 Years Old

A Case–Control Study

Abstract

Background: 

The significance of Clostridium difficile (CD) in the stools of children 2 years old or younger remains unclear. The aim of this study was to investigate risk factors and clinical evolution of diarrheic children ≤2 years old with or without CD in their stools.

Methods: 

From January 1, 2012 to December 31, 2013, all diarrheic stool samples received in our laboratory were screened for CD. We randomly selected diarrheic children ≤2 years old (n = 100) with an isolation of toxigenic CD in the stools and compared them with diarrheic children (n = 100) without isolation of CD.

Results: 

Cases and controls were appropriately matched for age and sex. We found no significant differences between children with or without CD. Of the CD cases, we compared the patients receiving treatment with metronidazole (19%) versus those that were not prescribed treatment (81%), and found that patients in the first group had used more gastric acid suppressants (P = 0.02), had surgery in the last month (P = 0.03) and also presented with more days with diarrhea (P = 0.03). All the patients, including CD cases, independently of the administration of metronidazole, were cured of the diarrheic episode. Polymerase chain reaction-ribotyping performed in all CD cases showed that the most prevalent ribotype was 014 (25%).

Conclusions: 

Our study reinforces the nonsignificance of CD in neonates and infants younger than 2 years old. Informing clinicians of CD isolates in this population promotes the use of antibiotics against CD, without evidence of a different outcome than those not receiving treatment.

Clostridium difficile infection (CDI) is known to be the most common health care-associated disease in adults,¹ mainly associated with risk factors, such as advanced age, duration of hospital stay, duration of antibiotic course and severity of underlying diseases.^2,3 CDI in adults causes elevated morbidity and mortality.⁴

The significance of the isolation of toxigenic C. difficile (CD) in children less than 2 years old is not well known.³ Up to 60% of children <24 months old are colonized with this organism,^5–8 thus trying to differentiate between true disease and simple carriage may be controversial.^9,10 Most actual microbiology laboratories do not search systematically patients less than 2 years old with diarrhea for the presence of CD.

In this context, we have sought to study children less than 2 years old with diarrhea and isolation of CD and compare results with children in the same age group with diarrhea but without CD isolation. Our institution screened for CD all diarrheic stools from all populations, including children, during the study period and informed clinicians of the results. The aim of this study was to investigate risk factors and clinical evolution of diarrheic children less than 2 years old with and without fecal isolation of toxigenic CD in their stools.

MATERIALS AND METHODS

Setting, Study Design and Population

Our institution is a 1550-bed general teaching hospital, including the maternity and pediatrics department that accounts for 350 beds. Our hospital attends a catchment population of approximately 715,000 inhabitants. The pediatric population represents an 18% of the whole population, of which 3% accounts for children less than 2 years old.¹¹

The present case–control study includes data from January 1, 2012 to December 31, 2013, in which all diarrheic stools received in our laboratory were screened for CD. The cases were identified randomly selecting 100 pediatric patients less than 2 years old with one or more diarrheic stool samples positive for CD. Each case of CD was matched with 1 control. The matching criteria were pediatric patients less than 2 years old presenting with a diarrheic stool sample, but in which the study for CD resulted negative.

Sources of Clinical Information

The patients included in the study were selected from a database of the microbiology department, in which all patients tested for CD were registered. Data in the database included age, gender, date of received sample and results of the microbiological tests performed in each sample. From the database, we selected the cases and controls that met our matching criteria. Clinical information was obtained from the electronic medical charts of the patients, which included date of birth, underlying medical conditions, type of acquisition of the diarrheic episode, Vesikari severity score for diarrhea, request of specific CD search by the clinician and other microorganisms isolated in the same stool sample. We also collected information regarding exposition to risk factors in the last month: antimicrobials, gastric acid suppressant use (including proton pump inhibitors and histamine-2 receptor antagonists) and surgery; as well as data of the current episode, we collected: presence of blood and mucus in stools, days with diarrhea, signs of abdominal distension and value of white blood count. Finally, we searched for clinical outcomes of the current episode, categorized as cured, clinical improvement or death and data regarding specific treatment for CD. We then selected the cases that received treatment for CD and the cases without treatment and compared data before treatment: Vesikari severity score for diarrhea, previous use of antimicrobials, gastric acid suppressants and surgery in the last month, as well as days with diarrhea. We also compared data after treatment for CD: clinical outcome including days of hospital stay, cure of current diarrheic episode and new onset of diarrhea at 8 weeks following the current episode.

Definitions

Definitions applied in our study for CDI were guided by the criteria of the European Society of Clinical Microbiology and Infectious Diseases (ESCMID) Study Group for Clostridium difficile¹² and the American Academy of Pediatrics.¹

Diarrhea was defined as 3 or more fluid or loose stools in 24 hours or more frequently than is normal for the individual. If samples did not qualify as diarrheic they were not processed for CD detection.

Community-acquired diarrhea: if initiation of the diarrheic episode occurred less than 48 hours after admission to the hospital and/or if the patient had not been discharged from a healthcare facility in the previous 12 weeks.

Healthcare facility onset: if initiation of the diarrheic episode occurred 48 hours after admission to a healthcare facility.

Healthcare facility associated: if initiation of the diarrheic episode occurred within the previous 4 weeks of the discharge from a healthcare facility or if the patient has constant contact with a healthcare facility, that is, patient receiving dialysis or chemotherapy.

Age groups were categorized as

• –Neonates: from new born until 28 days of birth.
• –Infants: from 29 days of birth until 2 years old.

Vesikari clinical severity score for diarrhea was used as a tool for measurement of severity of a diarrheic episode.¹³ This score takes into account parameters that include symptoms identified as important in the clinical presentation profile: diarrhea, vomiting, fever, dehydration, duration of diarrhea and vomiting and treatment status. Each of the 7 parameters is broken into thirds according to an equally divided severity distribution. The scores for each parameter within the clinical severity scoring system are added allowing for a severity score between 0 and 20 points. Severity scores above 15 points are considered severe, scores between 9 and 14 are considered moderate and scores less than 8 are considered mild.

Microbiologic Procedure

Screening tests were performed in diarrheic stools received in the microbiology laboratory, which was performed with C. Diff Quik Chek Complete (Techlab, Blacksburg, VA); in this test, a diagnostic algorithm was applied in which a GDH+/Tox AB+ result was considered positive and a GDH−/Tox AB− result was considered negative. When a GDH+/Tox AB− result was obtained, polymerase chain reaction was performed to detect toxin B gene (C. difficile GeneXpert, Cepheid).

Apart from the previous screening, other reference assays used to detect CD throughout the study period were a direct cytotoxicity assay and toxigenic culture, as previously reported.¹⁴

On all positive CD samples, toxinotyping was performed as previously described.¹⁵ Isolates were characterized by polymerase chain reaction-based ribotyping, with amplification of the 16S and 23S intergenic regions.¹⁶ The primer sequences were 59-GTG CGG CTG GAT CAC CTC CT-39 (16S primer) and 59-CCC TGC ACC CTT AAT AAC TTG ACC-39 (23S primer). Amplification involved 35 cycles of 1 minute for denaturation at 94°C, 1 minute for primer annealing at 57°C and 1 minute for extension at 72°C. Ribotyping patterns were separated by electrophoresis on MS-8 agarose (3.0%) at 100 V for 3 hours and analyzed with ultraviolet light after ethidium bromide staining. Phylogenetic analysis of ribotyping profiles was conducted with the unweighted pair group method with arithmetic mean and Dice coefficients (Bionumerics software 5.0, Applied Maths, Belgium). Ribotypes were named according to the international designation. When there was no correspondence with international ribotypes, the letter R followed by a number was used.

Data Analysis

Data were analyzed using the IBM SPSS software package, version 19.0 (IBM Corp, Chicago, IL). Qualitative variables for both CD cases and controls appear with their frequency distribution in absolute values and percentages. Continuous variables were reported as SDs (or median and ranges, where appropriate), with a Student’s t test nonparametric test to explore a possible association. Categorical variables were reported as proportions and 95% confidence intervals (CIs), with inferential analysis performed using χ² testing. Conditional logistic regression analysis was performed, yielding odds ratios (ORs) for the various parameters assessed. A level of significance of 0.05 was adopted for all inferential testing.

Ethical Issues

The Clinical Investigation Ethics Committee of Clinical Investigation of the Hospital General Universitario Gregorio Marañón approved the study titled “Clinical Significance of Clostridium difficile in Children less than 2 years of age: A Retrospective Matched Case-Control Study”. Approval code: micro.hgugm.2014-004.

RESULTS

During the study period, we included 100 cases and 100 matched controls. Baseline characteristics of case and control patients are offered in Table, Supplemental Digital Content 1, https://links.lww.com/INF/C337. Cases and controls were appropriately matched for age and sex (age group categorized as infants in 98%, 51% male for cases, vs. 94% infants, 64% male, for controls). There were no significant differences between the 2 groups regarding underlying medical conditions, type of acquisition of the diarrheic episode, Vesikari severity score for the diarrheic episode and request for CD in the stool sample or another microorganisms isolated in the same stool sample. The majority of patients in case and control groups did not have an underlying medical condition (77% vs. 75%), had a community acquired diarrhea (78% vs. 70%), had a mild Vesikari score for diarrhea (70% vs. 65%), did not have a CD request made by the clinician (89% vs. 93%) and did not have another microorganism isolated in the same stool sample (75% vs. 68%). The most common microorganisms found in both groups of patients were Rotavirus (11% vs. 11%) and Campylobacter jejuni (9% vs. 14%).

When studying risk factors for CDI, we did not find any significant differences between both groups of patients, respectively: previous antimicrobial use, 18% versus 22% (P = 0.48); previous gastric acid suppressant use (including proton pump inhibitors and histamine-2 receptor antagonists), 14% versus 10% (P = 0.38) and having a surgery in the last month, 5% versus 10% (P = 0.18; Table 1).

TABLE 1: Risk Factors for CDI and Current Diarrheic Episode of Cases and Controls

In relation to the current episode of diarrhea, we did not find significant differences between CD cases and non-CD controls regarding days with diarrhea (mean days 7 vs. 4, P = 0.11), presence of blood in stools (18% vs. 14%, P = 0.44), mucus in stools (23% vs. 25%, P = 0.74), abdominal distension (3% vs. 5%, P = 0.47) or mean white blood count (11,674 vs. 12,806, P = 0.36). No patient in the non-CD controls received specific treatment against CD versus 19 patients that were prescribed metronidazole in the CD cases. All the patients in both study arms were cured of the diarrheic episode (Table 1).

We sought to determine the reason for prescription of metronidazole and clinical outcome after administering treatment. Therefore, we studied only CD cases, and categorized patients who were prescribed treatment against CD versus patients who were not prescribed specific treatment. We found that patients who were prescribed metronidazole had used previously more gastric acid suppressants (32% vs. 10%, OR: 4.21, 95% CI: 1.25–14.14), had surgery in the last month (16% vs. 2%, OR: 7.40, 95% CI: 1.14–47.96) and also had presented with more days with diarrhea before prescription (mean days 16 vs. 5, OR: 1.05, 95% CI: 1.00–1.10). When comparing clinical outcomes after treatment, we found that patients receiving metronidazole had more days of hospital stay but this difference was not statistically significant (mean days 8 vs. 2 days, P = 0.15) and all the patients in the CD cases, independently of the administration of metronidazole, were cured of the current diarrheic episode (Table 2).

TABLE 2: Comparison Between Patients Receiving Treatment Against CDI and Patients Not Receiving Treatment in the Cases Group

Polymerase chain reaction ribotyping was performed on all toxigenic CD cases and results were as follows: the most common ribotypes found were 014 (25%), 017 (22%), 106 (18%) and 207 (10%); 9% of the ribotypes did not correspond with international ribotypes [R158 (2%), R123 (1%), R125 (3%), R94 (1%), R93 (1%) and R63 (1%)]; 7 ribotypes did not match with previous recorded ribotypes in our database, therefore were considered “new”; and 1 patient presented with 2 different ribotypes in the same sample (017 and 106; Table 3). No strain harbored the binary toxin-encoding genes; therefore, we did not find any 027 or 078 ribotypes among CD cases.

TABLE 3: Distribution of CD Ribotypes Among the CD Cases

DISCUSSION

Our study shows that there are no significant differences between diarrheic neonates and infants (<2 years old) with and without fecal isolation of toxigenic CD in their stools regarding underlying diseases, type of acquisition of diarrhea, severity of diarrheic episodes or isolation of other gastrointestinal microorganisms. We also did not find a significant relationship between risk factors for CDI that are common in adults and our studied population: previous use of antibiotics, gastric acid suppressant use or previous surgery.

Previous studies report that CD in neonates and infants is a colonizing agent rather than a pathogenic microorganism capable of causing diarrhea.^17–19 Some theories rely on the fact that the gut of neonates and infants lack the receptors needed to bind and process the toxins of Clostridium species.²⁰ Other theories report that the infant’s microbiota could provide an environment unfavorable to spore germination and thus absence of CD implementation,¹⁸ associated with a competitive intestinal colonization by nontoxigenic strains and toxin neutralization by maternal antibodies.¹ Contrarily to the above, other studies have reported CD to cause a wide spectrum of diseases in children from birth to adolescence, such as acute or chronic diarrhea and more rarely severe colitis.²¹ This seems to be related to preterm birth, neonates and children older than 3 years old. Thus, differentiation between true disease versus colonization of CD in children less than 2 years old has become controversial.²²

In a study by Rousseau et al,¹⁷ 294 infants (less than 2 years old) were prospectively screened for CD. They found CD in the stools of 99 infants (33%) and no case of CDI was diagnosed among carriers of a toxigenic strain. We found similar results, since diarrheic patients with CD did not differ from diarrheic patients without CD, suggesting that CD is not the cause of diarrhea in this population. Also, all of our studied patients had a favorable outcome with resolution of the symptoms and no further complications, independently of CD isolation.

A recent review of the literature regarding coinfection as a cofounder for CD in children concludes that this presentation may be common in children with diarrhea who test positive for CD and, therefore, a broad panel of pathogens should be tested to exclude other microbiological causes for diarrhea.²³ In our study, we found that the most common copathogens isolated in the same stool sample tested for CD were rotavirus and C. jejuni, in both study groups. However, we found no significant differences between diarrheic children with and without CD isolation and other microbiological causes for diarrhea.

The treatment for CD in children is based on adult guidelines, where metronidazole is recommended in the initial treatment of children and adolescents with mild to moderate disease and oral vancomycin is indicated as initial therapy for patients with severe disease and for patients who do not respond to oral metronidazole.^1,3 In neonates and infants, treatment constitutes a controversial decision, and most physicians decide not to treat children less than 2 years old.²⁴ In our study, 19% of the CD cases of children less than 2 years old were prescribed treatment with metronidazole. This probably occurred because these patients presented with more prior gastric acid suppressant use, surgery and more days with diarrhea. However, both groups of patients, independently of the prescription of metronidazole, were cured of the diarrheic episode.

Few studies have done molecular characterization of toxigenic CD in infants, and some also include nontoxigenic CD. In the same study mentioned previously by Rousseau et al,¹⁷ 98 CD strains isolated from infants were analyzed at a molecular level. Of them, 23 (7.1%) isolates were toxigenic: 21 were tcdA+/tcdB+ and 2 were tcdA−/tcdB+. They found a 15% of 014/020/077 genotypes, a 2% of 017 and a 1% of genotypes 048/012, 106 and 015. In our study, we found that frequent ribotypes isolated in adults, especially 014 and 106, were the most commonly found among our CD cases. Strains with these ribotypes have been associated to antibiotic resistance, particularly to quinolones, and to an elevated morbidity and mortality in this population.^25,26 Interestingly, ribotype 017, isolated in a 22% of our CD cases, has lately been increasing among the adult population.²⁷ Studies have reported this particular ribotype tcdA−/tcdB+ as causing epidemics of CDI in Asia and Germany.^28,29 In our study, we found a low prevalence of ribotype 001 (2%), frequently described in the adult population,^26,29 as well as no isolates with hypervirulent ribotypes 027 or 078, which have been related to severe illness and elevated mortality in adults.^30,31 These findings show that most strains found in infants are the ones commonly found in adults, suggesting that infants can be vehicles of transmitting CD to a more vulnerable population.

Our study presents the main limitation of being a retrospective study. Therefore, there are several data that we could not asses. As an example, information regarding breast feeding versus bottle feeding between both groups of patients was not available in all the electronic medical charts. However, our pediatrics hospital is keen in promoting breast feeding, so we could assume that most children included in the study, especially newborns, were initially breast fed. Another limitation is that our study was performed in a single center and during the study period we had no CD ribotype 027, a hypervirulent strain commonly found in other institutions worldwide.

We can conclude that our study reinforces the nonsignificance of CD in neonates and infants younger than 2 years old. Informing clinicians of CD isolates in this population promotes the use of antibiotics against CD, without evidence of a different outcome than those not receiving treatment. Therefore, microbiology departments should not systematically search for CD in children less than 2 years old. Ribotypes 014, 106 and 017 were the most prevalent in our study. No ribotypes 027 or 078 were found in our study group population.

ACKNOWLEDGMENTS

We would like to thank Mr. Lawrence Baron for his assistance with the English wording and final editing of the manuscript.