A comparison of five paediatric dosing guidelines for antibiotics

Abstract Objective To compare dosing guidance in the paediatric formularies of high- and middle-income countries for 32 commonly prescribed antibiotics on the World Health Organization’s (WHO’s) 2017 Model list of essential medicines for children. Methods We identified paediatric antibiotic guidelines that were either widely used internationally or originated from countries in which antibiotic use has increased markedly in recent years (i.e. Brazil, China, India, the Russian Federation and South Africa). Findings The study analysis considered five leading antibiotic guidelines: (i) the Manual of childhood infections: the blue book; (ii) the BNF (British national formulary) for children; (iii) the Red book®: 2018–2021 report of the committee on infectious diseases; (iv) WHO’s Pocket book of hospital care for children; and (v) Indian National treatment guidelines for antimicrobial use in infectious diseases. There was marked heterogeneity in the recommended dosing (i.e. daily dose, age dosing bands and dose frequency) for most commonly used antibiotics. The rationale for dosing recommendations was generally unclear. Conclusion The pharmacokinetic, pharmacodynamic and clinical evidence supporting paediatric antibiotic dosing, particularly on total doses and on age or weight dosing bands, needs to be improved. Future research should consider whether the variations in guidance identified stem from different clinical disease patterns, varying levels of antibiotic resistance or drug availability rather than historical preferences. Interested global parties could collaborate with WHO’s Model list of essential medicines antibiotic working group to develop an evidence-based consensus and identify research priorities.


Introduction
Global antibiotic consumption increased markedly between 2000 and 2010, with Brazil, China, India, the Russian Federation and South Africa accounting for 76% of the increase. 1 In response, the World Health Organization (WHO) developed a global action plan on antimicrobial resistance in 2015. 2 The fourth objective of this plan is to optimize the use of antibiotics. More recently, the classification of antibiotics in the WHO Model list of essential medicines has undergone substantial revision. 3,4 The new AWaRe classification divides antibiotics into Access, Watch and Reserve antibiotic groups with the aim of encouraging their rational use and optimizing prescribing. 3 Given this renewed focus on prudent antibiotic use, it is important that national prescribing guidelines are reviewed, any variations between countries are identified and the reasons for those variations are understood. Antimicrobials are the most commonly prescribed class of drugs for children. [5][6][7][8][9] Historically, however, paediatric dosing regimens have often been derived from pharmacokinetic data in adults, with the assumption that the relationship between drug exposure and total body weight is linear. 10,11 This approach, although clinically widespread, is not supported by solid empirical evidence and may result in neonates and children being exposed to inappropriate systemic drug levels. 12,13 The potential impact of inappropriate drug use in children on selection for antimicrobial resistance and the development of toxicity is unknown.
Clinicians' prescribing practices are often informed by formularies, which recommend antibiotic doses that balance efficacy, toxicity and drivers of antimicrobial resistance. However, recommendations are frequently based on historical practice rather than evidence. 14 For children, in particular, few data on efficacy, safety and pharmacokinetics are available. [15][16][17] Traditionally, there has been a preference for weight-based dosing strategies in the United States of America, whereas the United Kingdom of Great Britain and Northern Ireland has preferred age-banded dosing and WHO has preferred weight-banded dosing. This lack of standardization has resulted in widely varying recommendations and heterogeneous guidance, which have created ambiguities, especially for inexperienced clinicians. 18,19 While there has tended to be some agreement on adult dosing guidance, for example between the National Institute for Health and Care Excellence in the United Kingdom and the Infectious Diseases Society of America in the United States, this has not been the case for children. 20,21 Our view is that the international variation in guidance on paediatric dosing is most likely not based on different rates of antimicrobial resistance or clinical disease patterns, but instead reflects historical and cultural practices and the absence of a solid evidence base.
Although previous efforts have been made to compare local paediatric antibiotic guidelines, to the best of our knowledge there has been no detailed comparison of guidance from leading paediatric antibiotic formularies globally. 22 Consequently, the aim of our study was to compare antibiotic guidance in the paediatric antibiotic formularies of both highincome countries and emerging economies for 32 commonly prescribed antibiotics on the 2017 WHO Model list of essential medicines for children. 23

Methods
We identified antibiotic guidelines that were either widely used internationally or originated in countries in which antibiotic use has increased markedly in recent years (i.e. Brazil, China, India, the Russian Federation and South Africa). 1 In particular, we looked for guidance in these countries that had been endorsed by national and international bodies by contacting national coordinators of the Global Antibiotic Resistance, Prescribing and Efficacy among Neonates and Children (GARPEC) network. 24 Our aim was not to review all existing antibiotic guidance from every country comprehensively or to identify all patient management pathways. Instead, we selected guidelines that included specific paediatric dosing formularies or summaries, namely: Although we consulted national experts, we were unable to find paediatric antibiotic guidelines from Brazil, China, the Russian Federation or South Africa that were clearly endorsed nationally. From each publication, we obtained the recommended dosage of all antibiotics listed in section: 6.2 (i.e. antibacterials) of the 2017 WHO Model list of essential medicines for children, with the exception of: (i) benzathine benzylpenicillin; (ii) procaine benzylpenicillin; (iii) cefixime; (iv) tigecycline; (v) fosfomycin; (vi) daptomycin; (vii) polymyxins (e.g. colistin); (viii) fourth-generation cephalosporins, with or without a β-lactamase inhibitor (e.g. cefepime); and (ix) fifth-generation cephalosporins, with or without a β-lactamase inhibitor (e.g. ceftaroline). We included levofloxacin in our analysis even though it was not listed in the 2017 model list. We grouped the final selection of 32 antibiotics into the three AWaRe categories and compared dosing recommendations made by the different guidelines.
The BNF for children, in general, arranges guidance by route of administration and then by age band. The Blue book provides guidance by drug, giving the dose in milligrams per kilogram and the dosing frequency per day. The Red book® presents separate tables for neonates and children, with neonatal doses stratified by gestational age and then by postnatal age (where applicable): paediatric recommendations are given as a total daily dose per kilogram with a frequency of administration. The Pocket book of hospital care for children presents separate summary tables for neonates and children; guidance on neonatal dosing is given by route of administration and stratified by age into the first week of life and a postnatal age of 2 to 4 weeks. The Indian National Centre for Disease Control guidelines present two dosing summary tables: (i) a dosing guide for commonly used antimicrobial agents; and (ii) drug doses in the paediatric age group. We primarily used information from the first table, as the second table was less complete. However, we consulted the second table and the text to resolve inconsistencies. The Indian guidance did not present a separate table for neonates.
Where possible, we present doses in mg/kg per day or mg/day along with dosing frequency to enable guidelines to be compared. When dosing guidance was given for specific syndromes or suspected causative organisms, we included those for priority syndromes (i.e. pneumonia, sepsis, acute otitis media, pharyngitis and urinary tract infection) and for severe infection, especially meningitis. We excluded prophylactic doses, loading doses and doses for nonpriority syndromes and for low-or very-low-birthweight infants. In all cases where the guidance provide in summary tables was unclear, we consulted the primary text.

Results
For all 32 antibiotics studied, there were differences in recommended dosages across the five guidelines (Table 1 ; available at: http://www.who.int/bulletin/ volumes/98/6/19-234310). Of all the guidelines, the Pocket book gave dosing guidance for the fewest antibiotics reviewed (i.e. 12 of 32); it did not give recommendations for some commonly used Access (e.g. amoxicillin with clavulanic acid, phenoxymethylpenicillin and amikacin) and Watch (e.g. meropenem, piperacillin-tazobactam and vancomycin) antibiotics. In fact, it included guidance on only one antibiotic in the Watch group (i.e. erythromycin) and none in the Reserve group.
The number of dosing bands varied across guidelines, with the BNF for children tending to have the most and the Pocket book and the Indian National Centre for Disease Control guidelines having the fewest. The rationale for choosing different weight or age bands was not apparent, pharmacokinetic evidence was neither cited nor, if available, explained. 29 There was a considerable variation in recommended doses for each antibiotic. For example, Fig. 1 illustrates that the recommended dose of oral amoxicillin for a 5-year-old child weighing 18 kg with suspected nonsevere pneumonia varied from 360 mg per day (using the lower end of the Indian National Centre for Disease Control guidance of 20 mg/ kg per day) to 1620 mg per day (using the upper end of the Blue book guidance of 90 mg/kg per day). Moreover, the five guidelines all proposed different age ranges for dosing and the suggested dosing interval was 6, 8 or 12 hours.

Discussion
We found marked differences between the guidelines reviewed on paediatric dosing recommendations for 32 commonly prescribed antibiotics. There were differences in the age basis for dosing (e.g. the use of postnatal age alone or in combination with weight or gestational age), the frequency of administration and the total daily dose. Although some of this heterogeneity may reflect differences between settings in epidemiology, risk factors, causative organisms (e.g. due to different vaccination policies) or patterns of antibiotic resistance and may therefore be appropriate, other variations (e.g. the use of different age or weight dosing bands) can create confusion. In many instances, it is difficult to discern the rationale behind the dosing recommendations, whether they are derived from the accompanying summary of product characteristics, academic publications or expert consensus. Several national paediatric formularies, such as the Dutch, Italian and Spanish formularies, have been developed, but have been criticized for not presenting the underlying evidence or for referring to outdated paediatric dosing handbooks. 30,31 However, the Dutch formulary has clearly tried to address this issue. 32 Variations between guidelines may partly be due to a lack of robust evidence on the best treatment. In addition, defining optimal drug dosing is more complex for children than adults: antibiotic doses must be adapted to maturational changes in pharmacokinetics and consider changes in pharmacodynamics, yet still be simple and pragmatic. However, few studies on antibiotic pharmacokinetics or effectiveness have been performed in children. 17,33 Further, most of the guidelines we considered did not explicitly reference the evidence underlying their recommendations. The Blue book alone indicated the strength of the recommendations for each individual dose but did not present the supporting evidence. Moreover, the pharmacokinetic data underpinning many older off-patent antibiotics are very limited and were based on studies that were not conducted in accordance with recent standards. 34 A few strategic trials in children have examined high-and low-dose strategies and have reported efficacy and toxicity outcomes. However, these studies have used widely varying inclusion and exclusion criteria and end-points, which makes data synthesis between trials difficult. In addition, the absence of formal regulatory guidance on the design and conduct of antibiotic clinical trials in neonates and children has also hampered studies of clinical effectiveness. Our ability to compare therapies and treatment strategies would be improved by international collaboration to agree case definitions and outcome measures, similar to the collaboration that was effective in developing harmonized guidance for paediatric prescribing of antiretroviral drugs. 35

Dosing bands
The different age and weight dosing bands in leading global guidelines have been influenced by both practical considerations and historical usage. Although the most accurate method is to select the dose by weighing the child, this is not always possible. Simulations of orally administered amoxicillin in children show that the use of weight bands, rather than the exact weight, can often result in the administration of drug doses outside the therapeutic range. 14 Perhaps the simplest method of dose selection is to use age as a proxy for weight, though in practice this is often the least accurate method. Nevertheless, it can be useful when the child's recent weight is unknown, especially for drugs with a high therapeutic index (i.e. relatively safe drugs). Age bands need to be defined in a way that considers: periods of rapid weight change (e.g. during the first 6 months of life and around 8 to 9 years of age); the normal distribution of weight around the 50th percentile for specific ages; and local weight-for-age norms. 14 Although there are theoretical concerns that underdosing can increase the risk of treatment failure (the therapeutic level may not be reached) or promote resistance selection, these risks can be proved only in large clinical trials. On the other hand, children who receive too high a dose may experience adverse events and discontinue treatment. Antibiotic-associated diarrhoea is the most common side-effect of penicillin in children, with oral penicillin, the rate is increased in younger children and there appears to be an association with higher doses. 36 However, it remains unclear whether the child's age or the drug's formulation or dose is the main driver of severe diarrhoea. 36

Gaps in guidance
We identified several important omissions in global guidelines. First, we were unable to identify nationally endorsed, comprehensive treatment and dosing guidelines from Brazil, China, the Russian Federation or South Africa, although some individual institutions had their own local guidance. A unified approach within and, where appropriate, between countries would increase treatment standardization and make it easier to integrate the evidence. It is likely that many other countries, particularly lowand middle-income countries, also lack country-specific guidance and may have adopted recommendations developed elsewhere. In fact, published evidence often focuses on high-income settings, with little representation of low-or middle-income countries. 4 However, guidelines developed in one country may not be appropriate for countries where the epidemiology and burden of infectious diseases is different. Second, further research is needed into clinical syndromes in which the antibiotic dose must be altered to take account of vary-