Treatment of Community-Acquired Pneumonia: Are All Countries Treating Children in the Same Way? A Literature Review

Background Pneumonia represents an important threat to children's health in both developed and developing countries. In the last 10 years, many national and international guidelines on the treatment of pediatric CAP have been published, in order to optimize the prescription of antibiotics and limit their cost and side effects. However, the practical implementation of these guidelines is still limited. Main Text We analyzed the current recommendations for the therapy of pediatric community-acquired pneumonia (CAP) that all converge on the identification of aminopenicillins and beta-lactams as the optimal treatment for CAP. We also conducted a review of the current literature on antibiotic regimens used for pediatric CAP to identify the current state of guidelines implementation in different settings. We selected 37 studies published from 2010 to 2016, including both retrospective and prospective studies, mainly cross-sectional and hospital based. The results show a global heterogeneity in the antibiotics prescription for pediatric CAP, with application of guidelines varying from 0% to more than 91% and with important differences even within the same country. Conclusions Our review has demonstrated that the implementation of the guidelines is still limited but also that achieving the optimal prescription is possible and can be done in both developed and developing countries.


Introduction
Pneumonia is the single greatest cause of death in children worldwide, with an estimated 1.3 million deaths in 2011 and more than 90% occurring in developing countries [1][2][3]. It is responsible for 4% of deaths in newborns and 14% of deaths in pediatric patients [4]. The incidence of CAP is lower in developed countries: in the US it is about 35-40/1000/personyears in children < 5 years old, 20/1000 person-years in children 5-10 years old, and 10/1000 person-years in children > 10 years old. Despite this, approximately 50% of children with CAP < 5 years old, 20% between 5-10 years old, and 10% of children > 10 years old need to be hospitalized [5]. These numbers demonstrate the burden that CAP represents for society and for economic healthcare resources.

Materials and Methods
In the first part of the study, we compared the latest national and international guidelines on pediatric CAP, including all those who were published since 2005 to 2016, focusing on their recommendations for first-line therapies.
Then we performed a search on PubMed and Scopus databases, looking for studies published from 2010 to 2016 about CAP antimicrobial therapy in children, trying to get data from as many different countries as possible. We also performed hand-search of references of relevant articles. Our search included both retrospective and prospective studies, mainly cross-sectional and hospital based, including both inpatients and outpatients. All of them except for one [6] included pediatric patients only.
To get a more extensive review of CAP prescribing behavior, for those countries where specific studies on antimicrobial prescriptions for CAP were not available, a search for articles on antimicrobial prescriptions in pediatric age groups was performed. All articles including CAP as reason for treatment were included.

Different Countries, Same Pathogens.
Organisms responsible for CAP vary stratifying children by age because of the developing immune system and age-related exposures: viruses or mixed infections are more common amongst younger patients (children under 5 years of age), while exclusive bacterial origin and atypical etiology (mainly Mycoplasma pneumoniae and Chlamydophila pneumoniae) are more often identified in older children [7,8]. S. pneumoniae and Haemophilus influenzae are the commonest bacterial pathogens isolated in children under five years with CAP accounting for 30%-50% and 10%-30%, respectively [9]. Around 50% of deaths due to pneumonia are attributable to these organisms [10]. Viral etiology has been documented in up to 80% of CAP cases in children younger than 2 years and much less in older children (10-16 years). The most frequently identified viral pathogen in younger children is Respiratory Syncytial Virus (RSV), rarely detected in older children. Less frequent are Adenoviruses, Bocavirus, Human Metapneumovirus, Influenza A and B Viruses, Parainfluenza Viruses, Coronaviruses, and Rhinovirus. Up to 33% of hospitalized children are simultaneously infected by 2 or more viruses. Mixed infections (both of viral and bacterial etiology) have been documented in 2-50% of children with CAP, more frequently in inpatients, which are more seriously ill than outpatients [3,11].
Atypical pneumonia caused by different pathogens is characterized by a different clinical course: slowly progressing, with malaise, sore throat, low-grade fever, and cough developing over 3-5 days. The main organisms responsible for atypical pneumonia are M. pneumoniae in older children and C. pneumoniae in infants. Legionella species are rarely identified in children [8,12,13].
The etiologic definition is difficult for many reasons, such as low yield of blood cultures, difficulty in obtaining adequate sputum specimens from younger children, frequent specimen contaminations by upper airways bacterial flora and invasiveness of pulmonary biopsy, lung aspiration, and bronchoalveolar lavage which are rarely performed [13]. However, over the last 10 years, there have been improvements in PCR techniques for viral identification on nasopharyngeal aspirates or secretion, and molecular assays are now commonly used in Europe and in the US.
Vaccines are the most effective strategy for prevention of pediatric CAP. Haemophilus influenzae type B (HiB) conjugate vaccine and 7-valent pneumococcal conjugate vaccines (PCV7) dramatically decreased the incidence of bacterial CAP after introduction of universal vaccination campaigns [14,15]. PCVs have been included for some years in the immunization schedules of children in their first year of life in many countries and they have completely modified the burden of pneumococcal diseases among these children and their unvaccinated contacts of any age [16]. Currently, the polyvalent pneumococcal vaccine (PCV13) confers immunity to approximately 85% of serotypes responsible for most invasive pneumococcal diseases [17].

Same Pathogens, Same Treatment: International CAP
Recommendations. Since its introduction during the 20th century, antibiotic therapy, along with vaccines, has decreased CAP mortality of 97% in developed countries [14]. Most of the time the choice of an antimicrobial agent is empirical and based on the most common etiologies for each age group, on the local prevalence of causative organisms, and on the presence of risk factors for atypical or resistant bacteria [18].
During the last 10 years, many guidelines have defined the best antimicrobial regimen for CAP in children considering spectrum of activity, antimicrobial susceptibility, tolerability, bioavailability, safety, and cost [19,20]. As already highlighted by other authors, these guidelines present some differences in treatment strategies, but almost all agree on the first-line therapy to administer in case of CAP ( Figure 1) [19].
For infants < 2 months of age, the association with ampicillin and aminoglycosides is the most suggested therapy, ensuring coverage for Group B streptococci and Gramnegatives. In case of atypical pneumonia, in this period of life, because of the possibility of Chlamydia trachomatis infection, macrolides are recommended [3,19,[21][22][23].
For all children > 3 months of age, the narrowest regimen with S. pneumoniae activity is suggested worldwide. Penicillin is the ideal first-line therapy, being a narrow-spectrum agent achieving therapeutic concentrations for S. pneumoniae in the lung up to MIC of 4 mg/ml [24]. However, due to its limited bioavailability, oral amoxicillin is reported as an equivalent and more feasible option [24,25].
Despite general agreement on the agent, differences in dose and posology have been reported, varying according to pneumococcal resistance [19]. Indeed, beta-lactam effectiveness is time dependent and S. pneumoniae does not develop resistance through -lactamase enzyme production, but through the alteration of the cell wall's antimicrobial targets (penicillin-binding proteins) [26]. Thus, in the setting of resistant S. pneumoniae serotype, higher concentration at the infection site is needed in order to saturate penicillinbinding proteins and to overcome resistance [27].
A study of children with pulmonary pneumococcal infection [28] provided data to develop a model for describing amoxicillin pharmacokinetics administered with different patterns: 50 mg/kg/day in two or three administrations daily. The resulting curve, integrated with S. pneumoniae MIC for amoxicillin, showed that, for intermediate resistant S. pneumoniae (MIC 4 mg/ml) CAP, the amoxicillin plasma concentration remained above the pneumococcal MIC level for about 4 hours. Therefore, amoxicillin administered every 8 hours maintains blood and lung concentrations that are above S. pneumoniae MIC for enough time to allow S. pneumoniae eradication. A longer interval between administrations (every 12 hours), in case of intermediate resistant serotypes, would not permit having a sufficient antimicrobial plasma concentration [28]. Similarly, penicillin G needs more frequent administrations than other beta-lactams, because of its shorter half-life [13].
Beta-lactam dose is the other key factor for pathogen eradication. Through the different guidelines, amoxicillin daily dose varies from 40-50 mg/kg to 90-100 mg/kg, with higher dosage recommended in areas with higher risk for International Journal of Pediatrics antibiotic-resistant serotype, as in the US [13,19]. In the same way, for inpatient parenteral therapy, higher doses of penicillin G or ampicillin are recommended [13]. The only two guidelines which suggest an aminopenicillin plus beta-lactamase inhibitor as first line are the Taiwan Pediatric Working Group and Asociacion Espanola de Pediatria de Atencion Primaria [29,30]. Unlike the first one, in which aminopenicillin plus beta-lactamase inhibitor (e.g., amoxicillin-clavulanate) is suggested as first-line therapy for all children treated as outpatient, the Spanish guidelines recommend coamoxiclav only for children who are not fully immunized with conjugate vaccines for type B H. influenzae and for S. pneumoniae. Indeed, this population is at increased risk to develop a CAP by aggressive S. pneumoniae serotypes and other less common organisms, as H. influenza. Unlike Pneumococcus, type B and nontypeable H. influenzae became resistant to penicillin through the production of -lactamase. Therefore, treatment with the association of amoxicillin with a -lactamase inhibitor ensures a broader coverage [30]. It should be noted that the addition of alactamase inhibitor does not change the amoxicillin kinetic curve; as a consequence, in order to treat a pneumococcal infection with the association of amoxicillin with clavulanate, the therapy should be administered every 8 hours [26].
The WHO guidelines are the only one suggesting cotrimoxazole as alternative to amoxicillin in outpatient treatment. This recommendation derived from evidence of no difference in treatment failure rates between amoxicillin and cotrimoxazole [31][32][33]. Despite concerns about the increase of S. pneumoniae and H. influenzae resistant to cotrimoxazole, as demonstrated by some authors [34], the reason for this indication is mainly attributable to economic factors. Indeed, for children <10 kg, the cost of a five-day treatment with amoxicillin is higher than the same duration on cotrimoxazole [35][36][37].
No guidelines recommend oral cephalosporins as firstline therapy. Indeed, pharmacokinetic and pharmacodynamic studies showed that none of the available oral cephalosporins is able to exceed the pneumococcal MIC for more than 50% of the time between two administrations [26]. Moreover, recent US data on S. pneumoniae susceptibility to cefdinir and cefuroxime indicated only 70% to 80% efficacy, compared with 84% to 92% amoxicillin efficacy [38,39].
The only cephalosporin that has been demonstrated superior to penicillin in S. pneumoniae eradication, even if resistant, is ceftriaxone [40]. No microbiologic failures have been reported for S. pneumoniae with ceftriaxone MIC of 4.0 mg/mL [13,41]. Thus, ceftriaxone or cefotaxime in standard doses is suggested by all guidelines as alternatives in case of first-line treatment failure, severe clinical conditions, or not fully immunized children [3, 7, 13, 21-23, 29, 30, 41].
Due to high prevalence of macrolide resistance circulating strains of S. pneumoniae, macrolides are not recommended as empiric therapy for CAP. Their use is suggested only when atypical etiology is suspected or in case of persistence of symptoms despite beta-lactams administration [7,13,42]. This strict indication for macrolides use derives from the evidence that Mycoplasma lower respiratory tract infection (LRTI) has a high rate of spontaneous clinical remission and the use of azithromycin has been associated with the selection of resistant organisms because of its prolonged serum elimination half-life [13]. Moreover, no significant benefits of antibiotic treatment in M. pneumonia infection have been documented [37].
For complicated pneumonia (i.e., moderate parapneumonic effusion and necrotizing pneumonia), antimicrobial therapy must be broadened to cover less common but highly aggressive pathogens as Streptococcus pyogenes and S. aureus. As for S. pneumoniae, macrolides cannot be considered an effective empiric therapy because of the high level of resistance [13].
Despite the fact that no penicillin or cephalosporin resistance has been reported for S. pyogenes, some authors suggest that, in case of concomitant symptoms attributable to toxic shock syndrome, combination therapy with clindamycin decreases the severity of symptoms [43]. In fact, since clindamycin inhibits protein synthesis (by binding the 50S subunit of the bacterial ribosome), it inhibits the production of S. aureus toxins, resulting in a lower inflammatory reaction. Clindamycin may be bacteriostatic or bactericidal depending on the organism and drug concentration and is indicated by US guidelines as a good option for both methicillin susceptible S. aureus (MSSA) and communityacquired methicillin-resistant S. aureus (CA-MRSA) strains [13].
Nowadays almost all MSSA have penicillin resistance which can be overcome with the addition of a -lactamase inhibitor or through penicillinase-resistant beta-lactams, such as oxacillin or first-generation cephalosporins. MRSA strains have mecA gene that encodes penicillin-binding protein 2a, an enzyme that has low affinity for beta-lactams, leading to resistance to all antibiotics active against MSSA. During the last decade, both community-associated and hospitalacquired infections with MRSA have increased. MRSA, accounting for 20%-40% of all hospital-acquired pneumonia (HAP) and ventilator-associated pneumonia (VAP), have demonstrated a rapid increase as cause of pneumonia even in patients without exposure to the healthcare system [44]. This CA-MRSA has become an important cause of CAP complicated by empyema and necrosis [45].
Since erythromycin resistance predicts inducible clindamycin resistance in many isolates, a D-test to assess clindamycin susceptibility should always be performed. In case of D-test positivity, the use of clindamycin should be avoided, since it is highly possible that the organism will become resistant during the infectious process, especially in high-inoculum infections such as empyema [45]. On the other hand, all CA-MRSA strains are susceptible to vancomycin, which is considered by all guidelines as the drug of choice if MRSA is suspected [7,13]. Although linezolid has been recently demonstrated as efficient as vancomycin for the treatment of MRSA pneumonia, its use should be considered as a second-line treatment for cost consideration (linezolid costs >10 times more than vancomycin) and because linezolid-resistant MRSA has already been described [46,47].  Table 1 and Figure 2. Even if the studies were different in design and study population, their results give a good picture of the antibiotic prescription patterns in different environments, and they show the global heterogeneity in the application of the guidelines for the treatment of childhood pneumonia.
In fact, the first important result of our review is that the correct implementation of the guidelines is not confined to specific areas but may be variable even inside the same country. For example, Iroh Tam et al., through a 2-year retrospective study on hospitalized children with CAP in six US centres, showed that the most used antibiotics were thirdgeneration cephalosporins (73%), and only 1% of the patients received amoxicillin. These findings during the first 2 years after US guidelines publication led the authors to recommend more strategies for educating healthcare providers [71]. On the other hand, Thomson et al. in another retrospective study set in an US hospital, with the same population (hospitalized children between 3 months old and 18 years old) in a 15month period (May 2011-July 2012), had an opposite result, reporting that 63,6% of the pediatric CAP were treated with aminopenicillins and only 16.8% with third-generation cephalosporins [80].
Interestingly, in France our data about CAP prescriptions derive from two different settings. Launay and colleagues investigated antimicrobial prescriptions and recommendations adherence in a French Emergency Pediatrics Department through a prospective two-period study, including all children aged one month to 15 years. The results were encouraging, with an increase of recommendation compliance from 18.8% to 48% between 2009 and 2012, and a consequent increase of amoxicillin monotherapy prescription from 54.2% to 71% [78]. Dubos et al., on the other hand, give us a picture of CAP antimicrobial prescriptions through general practitioners (GPs), private pediatricians, and pediatric fellows. The results of the standardized questionnaire submitted to every participant showed that CAP guidelines were insufficiently followed, with high rate of amoxicillin/clavulanate prescriptions (amoxicillin in monotherapy was prescribed in only 29% of cases, for 54% of cases associated with clavulanic acid) [63].
In India, in addition, we found some of the lowest rates of prescription on aminopenicillins as single therapy. Choudry and Bezbaruah, in a prospective observational study based in a university hospital in Assam, including inpatients up to 12 years, reported 0% use of penicillin as single therapy in cases of pediatric pneumonia. The therapy mostly used (54% 6 International Journal of Pediatrics  of cases) was the combination of amoxicillin/clavulanate [58]. Another prospective study by Moinuddin et al. was conducted over 9 months in 2012, in two hospitals in Bangalore. The most widely used therapy was amoxicillin + clavulanate (43,8%), with third-generation cephalosporins as the most prescribed class (ceftriaxone 36.2%, cefotaxime 21%). Penicillin in single therapy accounted only for 1% of prescriptions [57]. Cephalosporins were often reported to be the class with higher rates of prescription for CAP treatment, as reported by many centres in different countries, like Ethiopia [60], Saudi Arabia [62], Nepal [73], Serbia [83], Sudan [67], US [54,71,74], Italy [76], and other European countries [48,82].
Feleke and colleagues conducted their 5-month prospective study in a large government hospital in Ethiopia. The study includes all children admitted in that period and CAP accounted for 56.3% of all drug prescriptions. Ceftriaxone was the most prescribed drug (43.5%) followed by gentamicin (25.6%), and penicillin and ampicillin ranked the third and fourth place [62]. In a retrospective study by Zec et al., during a 6-month period in 2014, first-and third-generation cephalosporins were given to children with CAP in 40.4% and 31.7% of cases, respectively. Penicillin was used in 25% of cases [83]. In an Italian 1-day point-prevalence survey on antimicrobial use in hospitalized neonates and children in 2012, the main indication for treatment in children was LRTI (34%), with higher prevalence of third-generation cephalosporins (43.3%) followed by macrolides accounting for 26.8%. No ampicillin/amoxicillin prescription was reported [76].
Association of aminopenicillins was found to be often prescribed: amoxicillin + clavulanate was reported to be the most used therapy by studies conducted in Saudi Arabia [51], France [63], and India [58], and a study conducted in Iraq, by Younis, reported that ampicillin + cloxacillin, alone and in combination, accounted for 50% of the antibiotic prescriptions for the children with respiratory tract infections [50].
One study, in particular, reported a high rate of prescriptions of macrolides. It was conducted in Norway, by Fossum and colleagues, and included the prescriptions of general practitioners in case of respiratory tract infections in patients < 6 years. They found that macrolides were prescribed in 44% of the cases of pneumonia, more than penicillin V, which was used in 31%, and that extended spectrum penicillin accounted for 24% of the prescriptions [55].
Studies on the appropriateness of prescriptions or prescriber behavior were also found. In addition to the aforementioned French study, Maltezou et al. showed how Greek private-practice pediatricians guidelines compliance is only around 30.6% [64]. Moreover, Ceyhan et al., in a multicenter point-prevalence survey with respiratory infection as main diagnosis, showed how cephalosporins and penicillin (most of the time combined with b-lactamase inhibitors) were improperly prescribed in 36.1% and 43.7% of cases, respectively. These analyses highlighted how, even now, adherence to guidelines is still low. On the other hand, Usonis and colleagues through a questionnaire developed and distributed by the CAP Pediatric Research Initiative (CAP-PRI) working group and distributed across Europe showed high adherence to CAP guidelines, with a high prescription rate of narrowspectrum penicillin for inpatients (amoxicillin (32%) and ampicillin (37%)) and outpatients (amoxicillin (84%)) [81].
An encouraging result is that almost a half (15/38) of the studies included in this review reported high rates of single therapy aminopenicillin or penicillin prescriptions. These studies were conducted in Brasil [75], Guyana [79], India [65], Mongolia [6], Nigeria [70], Tanzania [56], USA [80], Uganda [69], and France [78], showing that the current guidelines are applied in both developed and developing countries. The study by Awor et al. in Uganda in 2015 offers an important cause for reflection, since it shows that adherence to guidelines may be successfully implemented even in a nonhospital environment. In their 8-month quasiexperimental analysis, they investigated the visits and the prescriptions made by drug shop sellers, underlining how this class of health workers plays an important role in providing healthcare to populations in rural areas. Their result is that 91% of the children with pneumonia that were visited by drug shop sellers received amoxicillin, the highest rate of its prescription among all the studies included in this review [69].
Some data of antimicrobial prescriptions have been derived from point-prevalence surveys (PPS), including Australia [66,82], Mexico, Colombia, Argentina, Singapore, and European countries [48,49,81,82]. CAP was not the only analyzed disease, but the LRTI category was the most represented. Even though antimicrobial prescriptions were not specific only to CAP, PPS data were similar to the results of those other studies that were performed in the same country, but specifically designed for CAP.
Another interesting result is that the development of a local antimicrobial stewardship program could reduce inappropriate antimicrobial use and bacterial resistance, enhance patients' safety, and lower drug costs [84]. Moreover, global PPS could be a reliable and feasible tool for monitoring antimicrobial prescriptions all over the world.
Finally, it is also worthy of notice how data from certain countries were not available despite interest in the improvement of antibiotic prescription. For example, we did not find any report about pediatric CAP antibiotic treatment in Canada, even extending the research to 2005-2010. Likewise, we did not find any study set in other important countries, like China and Russia. It is worth remembering that the reduction of antimicrobial therapy and of microbial resistance is a global issue, and global effort is required in order to improve antibiotic prescription and administration practice.

Conclusions
In the last 10 years, many guidelines on the optimal treatment for childhood CAP have been published, with the aim of optimizing pediatric CAP antibiotic prescriptions. Our review demonstrates that the implementation of these guidelines is still limited but also that achieving the optimal prescription is possible and can be done in both developed and developing countries.  Hospital-acquired pneumonia VAP: Ventilator-associated pneumonia.

Additional Points
Availability of Data. The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Conflicts of Interest
The authors declare that they have no conflicts of interest.