Encouraging rational antibiotic use in childhood pneumonia: a focus on Vietnam and the Western Pacific Region

Globally, pneumonia is considered to be the biggest killer of infants and young children (aged <5 years) outside the neonatal period, with the greatest disease burden in low- and middle-income countries. Optimal management of childhood pneumonia is challenging in settings where clinicians have limited information regarding the local pathogen and drug resistance profiles. This frequently results in unnecessary and poorly targeted antibiotic use. Restricting antibiotic use is a global priority, particularly in Asia and the Western Pacific Region where excessive use is driving high rates of antimicrobial resistance. The authors conducted a comprehensive literature review to explore the antibiotic resistance profile of bacteria associated with pneumonia in the Western Pacific Region, with a focus on Vietnam. Current management practices were also considered, along with the diagnostic dilemmas faced by doctors and other factors that increase unnecessary antibiotic use. This review offers some suggestions on how these issues may be addressed.


Background
Childhood pneumonia is a major contributor to under-5 mortality, especially in developing countries [1,2]. In the Western Pacific Region, the highest burden of childhood pneumonia and pneumonia-related deaths occur in six countries: Cambodia, China, Laos, Papua New Guinea, the Philippines and Vietnam [3]. These countries report at least 0.2 pneumonia episodes per child year-more than 10 times the rates reported in developed countries from the same region such as Australia, New Zealand and Japan [4]. The management of childhood pneumonia is problematic in settings where a microbiological diagnosis is rarely pursued and the drug resistance profiles of common bacteria which cause respiratory disease are not readily available [5]. In many Asian countries this encourages excessive use of broad-spectrum antibiotics, irrespective of the child's disease severity. In the authors' experience, children are frequently hospitalized with pneumonia to administer intravenous antibiotics, despite relatively mild disease.
Unnecessary hospitalization for intravenous antibiotics increases healthcare cost, as well as treatment-related complications and the likelihood of nosocomial disease transmission. In order to encourage more rational use of intravenous antibiotics in children with communityacquired pneumonia, the authors performed a comprehensive review of common bacterial pathogens and their reported drug resistance profiles in the Western Pacific Region-Vietnam in particular. Factors that increase unnecessary antibiotic use (and measures that might reduce use) were also considered. PubMed, Google Scholar and Embase databases were searched using the following terms: antibacterial agents OR antibiotics OR drug therapy AND community acquired pneumonia OR acute respiratory tract infection AND child OR children OR childhood. Manuscript titles and abstracts were reviewed to identify original research papers that included children less than 5 years of age with pneumonia, with a geographic focus on Vietnam and the Western Pacific Region. In addition, the references of selected publications were reviewed, and co-authors suggested additional relevant papers. The term 'acute respiratory tract infection (ARTI)' describes all acute infections that involve the lungs or airways (upper and lower). The World Health Organization (WHO) defines 'pneumonia' as a child with tachypnoea, with or without signs of respiratory distress. Given the huge overlap in how these terms are applied in clinical practice, the authors considered 'pneumonia' and acute lower respiratory tract infection (LRTI) to be synonymous for the purpose of this review.

Common bacterial pathogens
Bacteria classically described to cause communityacquired pneumonia in children include Streptococcus pneumoniae, Haemophilus influenzae type b (Hib) and Staphylococcus aureus [6]. Table 1 presents an overview of pathogens commonly associated with acute LRTIs in children less than 5 years of age. With enhanced diagnostic tools, respiratory viruses and atypical bacteria such as Mycoplasma pneumonia are commonly detected in children with community-acquired pneumonia, particularly in studies from the Western Pacific Region [7][8][9]. This review focused on bacterial pathogens because they are the major cause of pneumonia-related mortality and the primary indication for antibiotic use. However, it is important to appreciate that accurate differentiation between viral and/or bacterial infection is a major challenge to clinicians and complicates management [10]. This review provides a brief overview of the most common bacterial pathogens and their reported drug resistance profiles from surveys conducted in the Western Pacific Region.

Streptococcus pneumoniae
The development of resistance mutations against multiple antibiotics is well documented in S. pneumoniae, as well as the expansion of resistant clones under antibiotic pressure. Resistance has been recorded against penicillin, tetracycline, cotrimoxazole, chloramphenicol, fluoroquinolones and macrolides [5,11,12]. Although the correlation between in vitro and in vivo resistance remains contentious, it is generally assumed that penicillin (at adequate dosages) should still be effective against pneumococcal pneumonia (not meningitis) caused by strains with low or intermediate levels of resistance in vitro [13]. This is difficult to verify, as many factors that influence the outcome of pneumonia treatment-such as underlying comorbid conditions, disease severity and supportive treatment-are poorly described in historical datasets and adequately powered head-to-head comparisons are rare [14][15][16]. Another important factor to take into consideration is the roll-out of pneumococcal conjugate vaccines. Routine administration of pneumococcal conjugate vaccine (PCV) in infancy has led to major reductions in pneumonia hospitalization and invasive pneumococcal disease in young children [17]. Given that the most common drug resistant strains were included in 7-and 13-valent pneumococcal conjugate vaccines, drug resistant pneumococcal disease has been greatly reduced in areas with high vaccine uptake [18], while reductions in strain carriage also reduced secondary pneumonia cases among older adults [17]. However, recent studies from Canada and the United Kingdom demonstrated substantial increases in multidrug resistant non-vaccine serotypes in both colonizing and invasive strains since introducing PCV13, with similar findings for colonizing strains after introducing PCV7 in Korean children [19][20][21].
High penicillin and macrolide resistance rates have been reported in published surveys, but many of these surveys included select patient populations and used minimal inhibitory concentration (MIC) breakpoints with variable stringency, and were also done prior to routine PCV delivery [5,22,23]. The use of variable MIC breakpoints in different studies is a major source of confusion and complicates study comparison. MIC breakpoints defined by the Clinical and Laboratory Standards Institute (CLSI) are widely used in the United States, while Europe has adopted the European Committee on Antimicrobial Susceptibility Testing (EUCAST) standards. They use broadly similar in vitro methods and specify MIC breakpoints by considering the pharmacokinetic-pharmacodynamic (PK-PD) properties of the drug, the clinical site of disease and the specific mechanism of drug resistance [24]. Studies have shown reasonable comparability between antibiotic susceptibility profiles using revised CLSI and EUCAST breakpoints  [25,26], but some significant differences remain. Against S. pneumoniae and H. influenzae, cefuroxime and cefaclor breakpoints still produce divergent results [26]. Global surveillance efforts would be greatly enhanced if uniform surveillance criteria can be agreed upon and MIC breakpoint definitions harmonized. A study conducted in 11 Asian countries reported a low (0.7%; 365/2184) prevalence of penicillin resistance in S. pneumoniae from non-meningeal isolates, [12] but it used an MIC breakpoint of ≥8 μg/ml as recommended by CLSI, which is much higher than the >0.06 μg/ml breakpoint proposed by EUCAST. A small case series of children with S. pneumoniae septicemia in China reported high drug resistance rates (96%; 24/25) [23], but MIC breakpoints were not specified. In Malaysia, 18% (5/28) of S. pneumoniae isolated from blood in children with community-acquired pneumonia was resistant to penicillin (using revised CLSI breakpoints) [27]. Since nasopharyngeal carriage may create a reservoir of resistant S. pneumoniae clones, it is important to monitor resistance in carriage specimens as well [28]. Nasopharyngeal S. pneumoniae carriage has been reported in 17% (102/614) of healthy Chinese children; 51% were resistant to macrolides by the E-test method [29]. In rural Vietnam, S. pneumoniae carriage has been detected in 50% of children, with higher rates in those less than 2 years of age (6-23 months). The resistance rate to amoxicillin and benzylpenicillin was low (4%; 17/ 421), based on revised CLSI breakpoints, but 95% were resistant to at least one antibiotic (cotrimoxazole 78%, erythromycin 70%, ciprofloxacin 28%) [30]. A more recent Vietnamese survey confirmed excellent penicillin susceptibility in S. pneumoniae isolates from respiratory specimens (penicillin 87% using revised CLSI breakpoints), although cephalosporin and macrolide susceptibility was poor (cefuroxime 19%, cefaclor 8%; azithromycin 4%) [31]. This supports the Vietnam national guidance to use high dose (90 mg/kg/day) amoxicillin as first-line treatment of community-acquired pneumonia in children [13].

Haemophilus influenzae
Although the prevalence of H. influenzae type b (Hib) has declined dramatically with widespread roll-out of conjugated Hib vaccine [32,33], it remains prevalent in settings with poor vaccine uptake. B-lactamase production is commonly associated with Hib infection. In Vietnam, 41% of respiratory Hib isolates produced βlactamase and 14% were β-lactamase non-producing ampicillin-resistant [31]. However, it is impossible to differentiate clinical infection from asymptomatic colonization using respiratory specimens from nonsterile sites. In China, Hib carriage decreased from 36% in 2000 to 19% in 2012, but β-lactamase-producing isolates increased from 4 to 31% over the same time period. Amoxicillin/clavulanic acid and 2nd or 3rd generation cephalosporins remained universally effective [34]. Despite the national Hib vaccine roll-out in Vietnam, Hib remains a common invasive pathogen [31], but this is expected to decrease as vaccine uptake improves. With increased vaccination uptake, the role of other encapsulated H. influenzae strains have increased [35]. A study in China found that 100% of H. influenzae strains identified in the sputum of children with pneumonia were non-typable; 1% (2/279) were β-lactamasepositive and 5% were β-lactamase non-producing ampicillin-resistant or intermediately resistant [36].

Staphylococcus aureus
S. aureus remains a common cause of communityacquired pneumonia. Pioneering lung puncture studies performed in Chile and Papua New Guinea identified S. aureus as a common pathogen in children with community-acquired pneumonia [37]. More recent studies found the pathogen predominantly in children at the severe end of the pneumonia disease spectrum, with increased frequency in severely malnourished children [38,39]. S. aureus also poses particular problems following influenza or measles infection, and as a secondary infection in hospitalized children [38,40,41]. A major challenge has been the emergence of methicillin-resistant S. aureus (MRSA) and more recently, a decrease in vancomycin susceptibility [42,43]. Among positive S. aureus blood cultures in Australian and New Zealand children, MRSA has been reported in 13% (142/1,073) with three times the average rate (incidence rate ratio, 3 [95%, CI: 2-4]), found among Aboriginal and Pacific Islander populations [44,45]. A recent Malaysian survey reported MRSA in 8% (3/38) of children with communityacquired S. aureus bacteremia; 32% had skin or soft tissues infections and 32% had community-acquired pneumonia [27]. A study assessing the prevalence of heterogeneous vancomycin-intermediately-resistant S. aureus (hVISA) in Asian countries (including South Korea, Taiwan, Hong Kong, Thailand, the Philippines, Vietnam, India and Sri Lanka) reported that among 462 MRSA isolates, 3.5% were hVISA, with the highest prevalence in South Korea and Vietnam (7.0%) [46]. A more extensive report on antibiotic resistance in 15 hospitals throughout Vietnam found MRSA in 20% of children and adults with S. aureus bacteremia; vancomycin resistance rates were very low [47].

Atypical bacteria
Mycoplasma pneumoniae is highly prevalent in children diagnosed with pneumonia in the Western Pacific Region [4]. M. pneumoniae is inherently resistant to all βlactams antibiotics and vancomycin, because it does not have a cell wall. With excessive macrolide use in recent years, reported macrolide resistance is near universal (90-100%) in parts of Asia [48]. Most countries report high rates of macrolide resistance (Japan 89% [49,50], China 83-98% [5,51,52], South Korea 63% [53], Hong Kong 47% [54], Taiwan 23% [55]), although the methods for M. pneumoniae susceptibility testing are poorly standardized. M. pneumoniae remains susceptible to macrolides in Australia where it is less frequently used [56]. Resistance to tetracyclines and fluoroquinolones have not been reported in clinical isolates and may provide a treatment alternative in some children, although reduced in vitro susceptibility has been reported [48]. However, most M. pneumoniae cases recover either without antibiotic treatment or despite documented drug resistance, thus the clinical value of antibiotic treatment remains uncertain [57].
Treatment of ARTIs is a major driver of antibiotic use in children. It is important for clinicians to be familiar with the most common bacterial causes, their local drug resistance profiles and the likely impact of antibiotic therapy (both positive and negative) in order to develop a rational treatment approach. Because a timely and definitive bacteriological diagnosis is currently impossible, empiric antibiotic treatment is justified in any acutely ill child. However, there is a need to critically consider factors that promote unnecessary and irrational antibiotic use, especially in children who are not acutely ill.

Factors promoting irrational antibiotic use
Antibiotic use generates selective pressure that increases the prevalence of drug resistant strains; hence, strategies to improve rational antibiotic use are important to protect antibiotics as a precious resource. A study of antibiotic use in Vietnamese hospitals showed that a large proportion of in-patients received inappropriate antibiotic therapy [58]. The main factors that promote unnecessary antibiotic use in the Western Pacific Region, using Vietnam as an exemplar, are listed below.

Unrestricted antibiotic access
Given unrestricted access to over-the-counter antibiotics, treatment of any respiratory infection with antibiotics is a common practice in Vietnam, Malaysia, and South Korea [59]. In Vietnam, the Pharmaceutical Law passed in 2005 requires an antibiotic prescription, but 38% of caregivers still access antibiotics without any formal medical assessment [60]; even injectable antibiotics can be acquired at local shops without prescription [61]. In the Western Pacific Region, antibiotic use before presentation to a doctor is common; more than 40% in Mongolia [62] and more than 50% of children admitted with ARTIs in the Philippines [63]. A study in nine international sites-Colombia, Ghana, India, Mexico, Pakistan, South Africa (2 sites), Vietnam and Zambia-found that the use of antibiotics in the 48 h prior to hospital admission was associated with treatment failure (OR: 1.8; 95% CI: 1.27-2.66) [64]. In addition, agricultural use of antibiotics remains essentially unregulated in most Asian countries, with high rates of colistin and cephalosporin use in the pig and poultry industries, resulting in increased rates of drug-resistant infections in human populations [65,66].

Unrealistic expectations and limited awareness
Limited awareness among the general public (including politicians) is a challenge in all settings; the WHO recently launched a program to increase awareness about antimicrobial resistance and the need for more prudent antibiotic use [59]. In response, country-specific strategies to limit antimicrobial resistance have been launched in the United Kingdom, the United States and Australia [67][68][69], but few Asian countries have followed suit. Unrealistic public expectation is a major factor driving excessive antibiotic use [59,70]. A study in rural Vietnam showed that just 13% of caregivers had correct knowledge about acute respiratory infections and 38% of caregivers self-managed common colds by buying antibiotics without prescription at the local pharmacy [60]. In Malaysia, 67% of people believed antibiotics to be effective against viral infections, with 47% using antibiotics during a common cold [61]. Moreover, antibiotic use is strongly influenced by cultural preferences and beliefs. Patients in Vietnam and China believe injectable antibiotics are more potent than oral options, and readily access injectable antibiotics without prescription [71]. A lack of adequate knowledge is also a problem among healthcare providers. A study in Vietnam demonstrated poor awareness about the risks and consequences of drug resistance among rural health-care providers when treating ARTIs; only 19% complied with recommended guidelines and 79% used antibiotics for common colds [72]. Table 2 provides an overview of physician-related factors that explains some of the excessive antibiotic use seen in Vietnam and parts of the Western Pacific Region. The difficulty in accurately differentiating viral from bacterial pneumonia is a major challenge in all settings. A study in Finland showed that of the routine blood tests, only the C-reactive protein (CRP) level differed significantly between bacterial and viral pneumonia patients [10]. Most children with dense lobar infiltrates on chest radiograph had laboratory evidence of a bacterial infection, but interstitial infiltrates were seen in both viral and bacterial pneumonia [10]. A recent randomizedcontrolled trial in Vietnam showed that point-of-care CRP testing reduced inappropriate antibiotic use for non-severe ARTIs [73]. A previous randomizedcontrolled trial in China investigated whether serum pro-calcitonin (PCT) could reduce unnecessary antibiotic use [74], but although the mean duration of antibiotic treatment was shorter in the PCT group, the antibiotic prescription rate was higher. It is hoped that rapid point-of-care testing for common respiratory viruses and biomarkers of severe bacterial infection will offer better bedside guidance in the near future, assisting more appropriate antibiotic use.

Physician-related factors
Because most child pneumonia deaths are caused by bacterial pathogens, current WHO guidelines recommend antibiotic use in all pneumonia cases, as defined by the presence of fast breathing. This approach may encourage the overuse of antibiotics, especially in children without danger signs who present with wheezing as a sign of reactive airway disease, which usually indicates a viral infection. Audible wheezing has been noted in 49% of Vietnamese children admitted with 'pneumonia' [75], which may identify a subgroup that does not require antibiotics [76]. WHO guidelines previously advised intravenous antibiotics in all children diagnosed with severe pneumonia (signs of respiratory distress), but oral amoxicillin was found to be effective in 92% (948/1,025) of children with a clinical diagnosis of severe pneumonia in Pakistan [77]. Subsequent trials demonstrated that home-based treatment can be applied to a wide variety of settings [74]. A multi-center study conducted in Bangladesh, Egypt, Ghana and Vietnam reported 9% (95% CI: 7-11%) treatment failure with 5 days of high dose oral amoxicillin (80-90 mg/kg/day), varying from 6% (95% CI: 3-10%) in Ghana to 13% (95% CI: 8-18%) in Vietnam [75]. A sizeable (but unknown) proportion of cases enrolled in these "clinical pneumonia" studies would have had viral infections. Therefore, it is not clear how much "treatment failure" actually resulted from inadequate antibiotic treatment for bacterial pneumonia; most children found to be "unresponsive to antibiotics" probably had viral pneumonia [78,79]. Despite the available evidence, most doctors in Vietnam routinely hospitalize children with "clinical pneumonia" to administer intravenous antibiotics; unnecessary hospitalization increases both healthcare cost and the risk of nosocomial infection.

Proposed recommendations
Actions and recommendations to improve rational antibiotic use in both the community and hospital environment are summarized in Table 3. Unrestricted antibiotic access and self-medication are firmly entrenched in Vietnam and most other Asian countries [80]. Given the multiple vested interests that protect the status quo, strong regulation and effective law enforcement will be required to limit excessive antibiotic use. In addition to public education programs, better training of healthcare providers to critically review the need for antibiotic use is essential [59]. Only 31% of countries in the Western Pacific Region report high awareness of antimicrobial resistance among their healthcare providers [59]. An educational program in Indonesia, including face-to-face clinician visits and group discussions, significantly Professional hierarchy • Junior physicians adopt the prescription habits of senior physicians without rigorous discussion or review of the evidence [86].
• In Vietnam, inappropriate antibiotic use is a particular problem in obstetrics, gynecology and surgery wards where professional hierarchy is most pronounced [58] No consideration of "societal risk" • Doctors and patients often prefer newer and more expensive antibiotics, which are considered more "powerful" [62] • Physicians provide antibiotics to help individual patients; potential societal risks are not considered [86]; • In the absence of functional microbiology services, physicians have limited information on local drug-resistance profiles and the impact of excessive antibiotic use; Perceived patient/ parent expectation • Doctors strive for patient satisfaction and if patients request antibiotics it is usually prescribed [70,87]; In Korea, 73% of doctors prescribe antibiotics for a common cold if requested by parents [88]; In Malaysia, 67% of patients believe that antibiotics help for viral infections [61] • Doctors have no time or motivation to explain the rationale for not using antibiotics Fear of poor patient outcome or litigation • Fear of poor patient outcomes is often listed as a key motivation for the use of broad-spectrum antibiotics by doctors [58,59,89] • Fear of litigation is not yet a major driver in the Western Pacific, but is likely to become a more prominent factor with increased development [90] Inadequate microbiology services • Near universal use of empiric broad spectrum antibiotics is common in places with poor microbiology services [70,87].
• In Vietnam, antibiotic use was reduced in hospitals with functional microbiology laboratories [58] Financial incentives to use antibiotics • Doctors' prescribing habits is influenced by personal income generated and incentives provided by pharmaceutical companies [70,86,87]. In China, as in many other Western Pacific countries, drug prescriptions supplement a doctor's income [62].
• In South-Korea drug dispensing by health care workers was banned in 2000, resulting in major reductions in antibiotic use [62] reduced the use of injectable drugs [81], but has not been replicated elsewhere. In 2013, the WHO facilitated a meeting of Western Pacific Region countries in Manila to identify feasible antimicrobial resistance (AMR) strategies [59], but this has not yet translated into revised childhood pneumonia guidance or management practices. In Vietnam, the Vietnam Resistance (VINARES) project aims to strengthen laboratory surveillance and reduce irrational antibiotic use. The project covers 16 participating hospitals [82], but outside of the participating hospitals AMR surveillance remains weak. Laboratory capacity to assist microbiological diagnosis and guide clinical management is insufficient throughout the region [83]. Functional antibiotic stewardship programs have been established in some countries, but this should become standard practice and be adapted to the local context [44]. An antibiotic stewardship program in Chinese hospitals set targets for antibiotic prescriptions and penalized doctors who prescribed antibiotics inappropriately [84]. After 2 years, antibiotic prescriptions decreased by 58-68% in inpatients and 15-25% in outpatients [85]. An antimicrobial stewardship program is currently being implemented in Vietnam (through VINARES), but the scope of the program is limited and there is a need for similar programs, including regular audits of antibiotic use, in every hospital [82].

Conclusion
Optimal child pneumonia management presents an opportunity to reduce excessive antibiotic use in the Western Pacific Region. However, encouraging the rational use of antibiotics requires education of healthcare professionals, facilitation of cultural change, improved clinical guidance and the establishment of functional microbiology laboratories to monitor disease etiology and drug resistance patterns, together with the removal of inappropriate incentives and effective enforcement of national regulations to restrict antibiotic use in healthcare and agriculture.  In addition, a detailed assessment of pneumonia case management should be conducted to understand clinical decision-making and provide more pragmatic guidance to clinicians in the field. Exemplars of comprehensive national strategies to address antimicrobial resistance and limit excessive antibiotic use include the United Kingdom, the United States and Australia [67][68][69]