Reported Rates of Diarrhea Following Oral Penicillin Therapy in Pediatric Clinical Trials

Pathophysiology of AAD

The spectrum of AAD ranges from mild symptomatic loose stools to life-threatening colitis.4 Unfortunately, very few studies have been performed on the mechanisms underlying AAD in children. As with adults, the most commonly cited mechanism for AAD is intestinal overgrowth with pathogenic microorganisms, especially Clostridium difficile, following alterations to the normal intestinal flora induced by antibiotics. However, in a study from 19965 that evaluated AAD and C difficile in stool specimens from children before and after a 10-day amoxicillin/clavulanate course for AOM, only 27% of children with amoxicillin/clavulanate–related AAD were positive for C difficile toxins on enzyme immunoassay; none of the affected children had C difficile toxins on enrollment. Other mechanisms for AAD are therefore hypothesized to include antibiotic-induced disturbance to the functions of the normal intestinal flora and to gut motility. These 3 mechanisms are described below.

Disturbance to Functions of Normal Flora

In adults 2 key disturbances have been described.4,6 Although it is likely that these also occur in children, further studies will be needed to verify this. First, various antibiotics (including ampicillin) have been reported to reduce the fermentation of carbohydrates by colonic bacterial flora. As a result there is an increase in undigested carbohydrates in the colon, leading to an osmotic diarrhea. Second, owing to the reduced carbohydrate fermentation, there is a deficiency in the short chain fatty acid (SCFA) metabolites from this process, which normally stimulate salt and water reabsorption in the colon. This lack of SCFAs can lead to profuse diarrhea.

Impact on Gut Motility

It is thought likely that the clavulanate component of amoxicillin/clavulanate stimulates peristalsis in children, leading to increased intestinal motility, and that a reduced clavulanate component will reduce diarrhea rates.7 This is discussed further in the “Drug and Dose” section below.

Overgrowth with Pathogenic Microorganisms

Antibiotics, especially penicillins, cephalosporins, and clindamycin, have been reported6 to alter intestinal flora and amino acid contents. As a result, microorganisms, including C difficile, Clostridium perfringens, Salmonella, Staphylococcus aureus, and Klebsiella, which are usually kept in check by a combination of bacterial antagonism and mucosal defenses, become pathogenic.6 In this regard, it has been reported8 that up to 70% of infants may be asymptomatically colonized with non-pathogenic, commensal C difficile but that this colonization rate decreases with age, reaching adult levels in children older than 2 years. However, recent studies8,9 have demonstrated that the US rate of pediatricC difficile–related hospitalizations increased from 7.24 to 12.80 from 1997 to 2006, with an increase in the severity of infection due to a new hypervirulent strain. Interestingly, an increasing proportion of children with C difficile infection (CDI) have community-associated disease, and with many of these infections there has been no history of antibiotic use. Importantly, while CDI symptoms usually develop 4 to 10 days after the initiation of antibiotic therapy, they can also take several weeks to develop after discontinuation of antibiotics.8

Literature Search and Study Selection

Our review focused on AAD in children arising from the use of oral preparations of amoxicillin, amoxicillin/clavulanate, flucloxacillin, and penicillin V (hereafter referred to as the “oral penicillins”). Using the Ovid databases, we conducted an Advanced Search in the EMBASE database (1980–March 28, 2012) and MEDLINE databases (Ovid Medline In-Process and Other Non-Indexed citations and Ovid Medline 1946–March 28, 2012) for any article reporting on rates of AAD arising from the use of any oral penicillin to treat an indicated infection in children (0–17 years). There was no restriction on the language of publication or the location of the reported studies. We then limited the articles to those reporting on clinical trials only.

Outcome Measure

The pre-specified outcome for data collection from each article was the rate of AAD arising from the use of an oral penicillin for an indicated infection. Additional data collected included the following: the dose and duration of the oral penicillin therapy, the definitions of diarrhea and diarrhea severity used, the age and size of the study population, the length of posttherapy follow up, and the number of children who discontinued therapy as a result of AAD (these items are collectively referred to as “relevant data” for the purposes of this review). Specifically, it was assumed that the articles used the World Health Organization (WHO) definition of diarrhea—“the passage of 3 or more loose or liquid stools per day, or more frequently than is normal for the individual”10—unless otherwise stated, and the Division of Microbiology and Infectious Diseases (DMID) severity grading set out in Table 1.11

Table 1. DMID Severity Grading for Diarrhea in Children Older Than 3 Months of Age

View Fullscreen

Data Screening and Collection

The title and abstract of all articles were screened by 2 independent researchers for relevant data or for a reference indicating relevant data were available in the full text. The relevant data for each article were then extracted onto a data collection form. Any article that primarily focused on diarrhea occurring in children with chronic conditions (e.g., cochlear implants in otitis media studies) or that did not offer information about the number of pediatric patients or the dose of oral penicillin given was excluded. Likewise, articles reporting on the concurrent use of any oral penicillin with other antibiotics (e.g., in the treatment of Helicobacter pylori) were excluded. On completion of the data collection process the researchers agreed upon a list of articles for inclusion in the primary analysis.

Data Analysis

Primary analysis involved the preliminary pooling of studies and calculation of AAD rates for each of the oral penicillins. To calculate this rate, the total number of children reported in the studies to be affected by diarrhea from each of the oral penicillins was divided by the total number of children receiving that oral penicillin. No corrections were made to the AAD rates reported as an adverse event in the studies. Specifically, patients who withdrew from the studies were not excluded.

Limitations and Potential Bias

Only 1 of the articles reviewed38 indicated the diarrhea definition used in the Study. Hoberman et al38 stated that “protocol-defined diarrhea … is the occurrence of three or more watery bowel movements in 1 day (1 day of protocol-defined diarrhea), or two watery bowel movements per day for 2 consecutive days (2 days of protocol-defined diarrhea).” Another Study27 confirmed that “diarrhea was not specifically defined in the protocol and information was collected as presented by parents or caregivers.” Otherwise there was reference to the “protocol-defined diarrhea” by Bottenfield et al34 and to the “Otitis Parent Questionnaire” by Block et al,21 but no specific definitions. Therefore, for the majority of studies it is not known what definition of diarrhea was used, if any. It was therefore assumed that, in the absence of a distinct definition, any reported diarrhea would meet the WHO definition of diarrhea stated above.10 However, it is known that the diarrhea was reported by parents and/or elicited by investigators and that the investigators determined whether the diarrhea was related to the oral penicillin (i.e., whether it was AAD or not). Unfortunately, none of the studies report on the timing or duration of AAD, although some studies confirmed that the AAD resolved upon discontinuation of the oral penicillin. In addition, as AAD can occur up to 2 months after the final antibiotic dose, there is the risk of underestimation of AAD where follow up in the studies occurred for a shorter period of time.50 As shown in Tables 2 through 4, the length of follow up in the studies ranged from 1 to 46 days post–completion of therapy, with less than 50% of the amoxicillin/clavulanate studies having posttherapy follow up of 28 days or more.

In the majority of the studies the investigators categorized diarrhea, along with other adverse effects, as being mild, moderate, or severe, but again, no definitions were stated for this severity grading. It was therefore assumed that the DMID severity grading (Table 1) was used.11 Overall, 55 cases of AAD were reported as severe or leading to withdrawal from a study, and there were no reports of any deaths or life-threatening colitis.

Drug and Dose

Penicillin V is a broad-spectrum penicillin similar to the original benzylpenicillin or penicillin G. It was introduced because it was gastric-acid stable and could therefore be given orally (British National Formulary).51 It is principally indicated in the acute treatment of streptococcal tonsillitis, pharyngitis, and scarlet fever and has the additional benefit of preventing long-term sequelae of these streptococcal infections, such as rheumatic fever and glomerulonephritis.46 Only 3 of the studies reported on diarrhea rates following the use of penicillin V (Table 4). Cohen et al16 reported that only 1 of 160 penicillin V–treated patients (0.63%) was affected by AAD. Pichichero et al46 reported that 4 out of 202 penicillin V–treated patients (2%) were affected by AAD. The third study by Block et al47 reported that none of 55 penicillin V–treated patients were affected by AAD. These minimal rates accord with the 3% (2 patients out of 59) combined penicillin V and penicillin G–related AAD rate reported by Turck et al49 in their epidemiological study of the incidence of oral antibiotic–associated diarrhea in an outpatient pediatric population. Therefore, although only 3 studies have been compared, the low rate of penicillin V–related AAD identified appears to be consistent with the findings of other studies.

Amoxicillin is a semisynthetic broad-spectrum penicillin that was introduced in the United States in 1974.7 Like all penicillins, its antimicrobial activity is due to its inactivation of bacterial cell wall synthesis via its beta-lactam ring structure. As it was better absorbed and had higher serum concentrations than ampicillin (introduced in the United States in 1962) it soon became, and today remains, the first-line therapy for uncomplicated AOM, sinusitis, and mild to moderate pneumonia in children. In terms of AAD, our results show that use of amoxicillin resulted in a pooled diarrhea rate of 8.1%, with rates ranging from 1.87% to 17.5% (Table 2). This range is consistent with the AAD rates ranging from 3% to 16% found by Coker et al2 in their review of amoxicillin-related diarrhea rates reported by clinical trials. Finally, our data analysis found no correlation between the dose of amoxicillin and rate, which may well have been due to the small number of amoxicillin studies.

Amoxicillin/clavulanate was introduced in the United States in 1984 to combat the beta-lactamase–producing bacteria that were inactivating amoxicillin (particularly strains of the key causative bacteria of pediatric AOM, sinusitis, and pneumonia: Streptococcus pneumoniae, Haemophilus influenzae, and Moraxella catarrhalis).7 Amoxicillin/clavulanate is a combination of amoxicillin with clavulanic acid, a beta-lactamase inhibitor. Its original formulation combined amoxicillin/clavulanate in a 4:1 ratio (40/10 mg/kg/day).7 However, as it was realized that effectiveness against drug-resistant S pneumoniae required not only a beta-lactamase inhibitor but also increased concentrations of amoxicillin at the site of infection, the amoxicillin component of amoxicillin/clavulanate was increased to a 7:1 ratio (45 mg/6.4 mg/kg/day) and to an 8:1 ratio (80 mg/10 mg/kg/day) and then, in 2001, to a 14:1 ratio (90 mg/6.4 mg/kg/day).7 As a result of its effectiveness against amoxicillin-resistant strains of bacteria, amoxicillin/clavulanate has become a treatment of choice for children who fail initial therapy with amoxicillin.7 In terms of AAD, our results show that use of amoxicillin/clavulanate resulted in a pooled diarrhea rate of 19.8%, with rates ranging from 6.7% to 47.8% (Table 3). Again, this is in line with the findings of other reviews: Coker et al2 reported a 20% amoxicillin/clavulanate diarrhea rate following their systematic review. It is therefore clear that this combined formulation causes considerably more AAD than does amoxicillin alone.

However, there is an interesting trend within these data. When analyzing the formulations used in the amoxicillin/clavulanate studies for the treatment of AOM we found that there was a pooled average diarrhea rate of 24.6% for the 4:1 formulation, a rate of 12.8% for the 7:1 formulation, a rate of 19.0% for the 8:1 formulation, and a rate of 20.2% for the 14:1 formulation. These rates are recorded in Table 5. A proportional reduction of clavulanate from 4:1 (40 mg/10 mg/kg/day) to 7:1 (45 mg/6.4 mg/kg/day) in the amoxicillin/clavulanate formulation appears to correlate to a comparative reduction in amoxicillin/clavulanate–related AAD; however, this is not maintained when the amoxicillin component is doubled from 4:1 (40 mg/10 mg/kg/day) to 8:1 (80 mg/10 mg/kg/day) or from 7:1 (45 mg/6.4 mg/kg/day) to 14:1 (90 mg/6.4 mg/kg/day). This initial trend to reduction was previously identified by Hoberman et al38 in their study comparing the 4:1 (26.7% AAD rate) and 7:1 (9.6% AAD rate) formulations. One explanation offered by Klein7 in his review of amoxicillin/clavulanate therapy for pediatric infections is that clavulanate stimulates peristalsis; hence, reducing its concentration will reduce any AAD. However, this fails to explain the apparent increase in diarrhea associated with the 14:1 current formulation. It has also been reported by Harrison52 that the rates of amoxicillin/clavulanate–related AAD can be reduced by administering amoxicillin/clavulanate with food, as bowel peristalsis is inhibited after eating. It is nevertheless clear that amoxicillin per se causes AAD and that increasing the amoxicillin component of amoxicillin/clavulanate will increase AAD, notwithstanding a relative reduction of clavulanate, hence the 19.0% and the 20.2% AAD rates identified for the 8:1 and 14:1 formulations, respectively.

Table 5. Range and Averages of Antibiotic-Associated Diarrhea (AAD) Rates for Common Formulations of Amoxicillin/Clavulanate

View Fullscreen

Disease

The majority of the studies (34 of the 42) involved the use of amoxicillin/clavulanate (n = 30) or amoxicillin (n = 4) to treat AOM. Five studies involved the use of amoxicillin (n = 2) or penicillin V (n = 3) to treat streptococcal pharyngitis. Two studies used amoxicillin/clavulanate to treat pneumonia, and 1 study used amoxicillin/clavulanate to treat sinusitis. Therefore, in view of the dominance of the use of amoxicillin/clavulanate to treat AOM it was not possible to perform any meaningful analysis related to whether a particular disease condition is associated with a higher incidence of AAD. Furthermore, we are not aware of any study that has analyzed this as a possible risk factor for AAD.

Age

Another major potential risk factor is age. Turck et al49 studied AAD from penicillins G and V, penicillins A and M, amoxicillin/clavulanate, cephalosporins, macrolides, trimethoprim/sulfamethoxazole, and erythromycin/sulfafurazole and confirmed that the incidence of AAD is significantly greater in children who are <2 years of age compared to in those who are over 2 years of age. Specifically, they reported AAD rates of 18% for children aged 1 month to 2 years, 4% for those aged 2 to 7 years, and 2% for those older than 7 years. Although the AAD age incidence was not reported for each type of penicillin, it was reported that the relative risk of AAD for a child receiving amoxicillin/clavulanate was 2.43 and that if the child was aged less that 2 years that relative risk increased to 3.5. It was not possible in our review to break down the AAD rates by age, as AAD rates were reported for too wide an age range in the studies.

Duration

The duration of antibiotic therapy is another possible risk factor. Some studies17,38 have demonstrated that a longer duration of amoxicillin/clavulanate is associated with an increased rate of AAD. Because very few of the studies had an oral penicillin duration of less than 10 days, we were unable to confirm this in our review.

Management

In view of the incidence of AAD with the oral penicillins and the increased incidence of community-associated CDI, parents should be advised of the relevant AAD incidence rate for the oral penicillin prescribed and advised to come back if their child develops moderate or severe diarrhea or becomes dehydrated. The management of AAD will depend on its severity. If it is not severe (i.e., if there are no features of antibiotic-associated colitis, characterized by abdominal cramps, fever, leukocytosis, fecal leukocytes, hypoalbuminemia), then discontinuation of the antibiotic or switching to an antibiotic with reduced risk of AAD should lead to its resolution in the majority of cases. In addition, electrolytes and fluids should be replaced. If the AAD is severe, then in addition to the measures for non-severe diarrhea, hospitalization will be necessary (with isolation measures to prevent spread), and a stool sample should be sent for C difficile toxin analysis. If the sample is positive for the toxins, then the CDI should be treated with oral metronidazole or vancomycin.

Prevention

The most effective way of preventing AAD is to reduce the inappropriate use of antibiotics. As this review has demonstrated, there is a significant risk of AAD following the use of amoxicillin and amoxicillin/clavulanate in AOM. It was expected that the introduction in the 2004 AOM clinical practice guideline of the American Academy of Pediatrics (AAP) and American Academy of Family Physicians (AAFP) of an observation period for mild cases of AOM in a select group of children before deciding if antibiotics are needed would reduce antibiotic use.1 However, the most recent 2013 AOM clinical practice guideline of the AAP and AAFP53 reported that while multiple studies had shown the success of an observation period, there had been no difference in the antibiotic prescribing rates since the 2004 guideline, and this was attributed to lack of clinician awareness. In addition, the most recent 2011 studies of Hoberman et al54 and Tahtinen et al19 underscored the importance of correctly diagnosing AOM cases requiring antibiotics after finding that young children with a certain diagnosis of AOM recover more quickly after initial treatment with amoxicillin/clavulanate versus observation (as identified by Klein55 in his editorial for these 2 studies). In order to address these issues the 2013 AOM clinical practice guideline set out recommendations for diagnosing and managing AOM in children aged 6 months and above. For diagnosis, the key recommendations include the following: 1) AOM should be diagnosed in children who have (a) moderate to severe bulging of the tympanic membrane or ear drainage not caused by otitis externa or (b) mild bulging of the tympanic membrane and less than 48 hours of new ear pain or intense redness of the tympanic membrane; and 2) AOM should not be diagnosed in children who do not have middle ear fluid. For management the 2013 guideline confirmed that clinicians should prescribe antibiotics for children 6 months and older with severe unilateral AOM and for children with non-severe bilateral AOM. However for non-severe unilateral AOM in children aged 6 months to 12 years the clinician can offer either antibiotics initially or observation. Under this option antibiotics (amoxicillin or amoxicillin/clavulanate) may be withheld initially and the child must be closely observed. If symptoms worsen within 72 hours, then antibiotic therapy should be started. The decision of whether to start antibiotics or observe will depend on the clinician's certainty about the diagnosis, the child's age, any comorbid conditions, the severity of the AOM, and the likelihood that the parents will follow up if the child's symptoms worsen or do not improve in 48 to 72 hours. Certainty of diagnosis, older age, absence of comorbid conditions, mild AOM, and reliable parents will make patients candidates for observation. It is expected that this observation period will reduce antibiotic use.

Once antibiotic therapy has begun, recent studies have looked at whether concurrent use of probiotic formulas has reduced the incidence of AAD. A recent 2011 Cochrane review56 of the use of probiotics for the prevention of pediatric antibiotic–associated diarrhea concluded that high-dose Lactobacillus rhamnosus and Saccharomyces boulardii of 5 to 40 billion colony-forming units/day may prevent the onset of AAD with no serious side effects in otherwise healthy children. However, it qualified this conclusion by suggesting that this benefit of high-dose probiotics be confirmed in a large randomized controlled trial. It also advised that probiotic use be avoided in children at risk of adverse events until further research has been carried out, as moderate to serious adverse events had been reported in severely debilitated or immunocompromised children with underlying risk factors, including central venous catheter use and disorders associated with bacterial/fungal translocation.

Finally, especially in view of reported increases in community-associated CDI, infection control measures to prevent the spread of possible CDI-related AAD will need to be emphasized when prescribing oral penicillins. Specifically, in the event of AAD, parents must be advised to ensure rigorous hand washing with soap by all family members and household cleaning with strong disinfectants to rid the environment of any C difficile spores.