Safety of azithromycin in pediatrics: a systematic review and meta-analysis

Purpose To evaluate the toxicity of azithromycin in neonates, infants, and children. Methods A systematic review was performed for relevant studies using Medline (Ovid), PubMed, Cochrane Central Register of Controlled Trials, EMBASE, CINAHL, and International Pharmaceutical Abstracts. We calculated the pooled incidence of adverse drug reactions (ADRs) associated with azithromycin based on prospective studies (RCTs and prospective cohort studies) and analyzed the risk difference (RD) of ADRs between azithromycin and placebo or other antibiotics using meta-analysis of RCTs. Results We included 133 studies with 4243 ADRs reported in 197,675 neonates, infants, and children who received azithromycin. The safety of azithromycin as MDA in pediatrics was poorly monitored. The main ADRs were diarrhea and vomiting. In prospective non-MDA studies, the most common toxicity was gastrointestinal ADRs (938/1967; 47.7%). The most serious toxicities were cardiac (prolonged QT or irregular heart beat) and idiopathic hypertrophic pyloric stenosis (IHPS). Compared with placebo, azithromycin did not show increased risk ADRs based on RCTs (risk difference − 0.17 to 0.07). The incidence of QT prolonged was higher in the medium-dosage group (10–30 mg/kg/day) than that of low-dosage group (≤ 10 mg/kg/day) (82.0% vs 1.2%). Conclusion The safety of azithromycin as MDA needs further evaluation. The most common ADRs are diarrhea and vomiting. The risk of the most serious uncommon ADRs (cardiac-prolonged QT and IHPS) is unknown. Electronic supplementary material The online version of this article (10.1007/s00228-020-02956-3) contains supplementary material, which is available to authorized users.


Introduction
Azithromycin is an acid-stable orally administered macrolide antimicrobial drug, structurally related to erythromycin [1]. Due to its broad antibacterial spectrum against Streptococcus pneumonia, Moraxella catarrhalis, and atypical pathogens, azithromycin has been used extensively for the treatment of pediatric infectious diseases and became one of the most commonly prescribed antibiotics in children [2][3][4][5][6]. During the last 20 years, azithromycin mass drug administration (MDA) has been used to control trachoma with over 700 million doses of azithromycin being prescribed to children in areas of active trachoma programs [7]. Recent large trials have suggested that periodical azithromycin MDA may reduce post-neonatal infant and child mortality [8]. However, the long-term rationale for mass antibiotic distribution for trachoma is still the subject of debate with concerns of potential toxicity with azithromycin in pediatrics [9,10].
A systematic review that evaluated the tolerance or toxicity of azithromycin in children with asthma found that gastrointestinal adverse reactions such as nausea, diarrhea, and abdominal pain were the main adverse events [11]. Another systematic review of azithromycin use in neonates highlighted the risk of infantile hypertrophic pyloric stenosis (IHPS) [12]. This systematic review aims to evaluate the toxicity of azithromycin both as MDA or non-MDA in neonates, infants, and children from birth to 18 years old. This systematic review was proposed by the World Health Organization (WHO), as one of the systematic reviews in support of developing a guideline of azithromycin use in pediatrics to help national and international policymakers in determining the role of prophylactic azithromycin in reducing child mortality [10].

Methods
This systematic review conformed to the PRISMA statement and was registered on PROSPERO (CRD 42018112629) [13]. We have reported the methods of literature search, risk of bias assessment, data abstraction, and data analysis in the published protocol [14].

Search strategy and literature search
In brief, we performed a comprehensive search using Medline (Ovid), PubMed, Cochrane Central Register of Controlled Trials, EMBASE, CINAHL, and International Pharmaceutical Abstracts. We reported the search strategies for each database in the published protocol [14]. We included randomized controlled trials (RCTs), cohort studies, case-control studies, crosssectional studies, case series, and case reports that included pediatric patients (aged from birth to 18 years old) using azithromycin as periodic MDA or as therapeutic agent for any disease till March 2019, and updated the search in September 2019. We had no restriction on language. We excluded editorials, conference abstracts, and reviews. We searched for adverse drug reactions (ADRs) reported in spontaneous reporting systems or safety communication announcements as planned. However, since none of the spontaneous reporting systems provided detailed information for individual ADRs (e.g., age of patient, dosage of azithromycin) and none of the announcements was based on evidence in patients younger than 16 years old, we did not present the result in this paper.

Data extraction and analysis
We abstracted study design, characteristic of patients, interventions, methods used for safety monitoring, and ADRs from the eligible studies. We used the Cochrane risk of bias tool to assess risk of bias in RCTs, the Newcastle-Ottawa Scale for casecontrol study and cohort studies, and the Joanna Briggs Institute (JBI) Critical Appraisal tools for case series, case reports, and analytical cross-sectional studies [15][16][17]. We used the Grading of Recommendations Assessment, Development and Evaluation (GRADE) to assess the certainty of body of evidence [18]. We categorized ADRs by systems according to the Medical Dictionary for Regulatory Activities (MedDRA) Terminology (version 21.1) [19] and by frequencies according to the Council for International Organization of Medical Science (CIOMS) as very common (≥ 10%), common (≥ 1% and < 10%), uncommon (≥ 0.1% and < 1%), and rare (< 0.1%) [20]. When the primary study explicitly reported zero ADR or when the study had the capacity to detect the ADR but did not report any event, we counted it as a zero event [21].
We calculated the pooled incidence of ADRs associated with azithromycin based on RCTs and prospective cohort studies to determine the risk of individual ADRs [22,23]. We categorized ADRs into those that needed specific investigations (e.g., "decreased white blood cells" needs to be detected by a blood test; "prolonged QT" needs to be detected by an electrocardiograph (ECG)) and those that could be observed without specific investigations (e.g., diarrhea, vomiting). We analyzed risk difference (RD) of ADRs between azithromycin and placebo or other antibiotics using meta-analysis of RCTs. We did not use relative risk (RR) or odds ratio (OR) as planned in the protocol, as one cannot use inverse-variance methods to calculate RR or OR, when the number of events is zero in either group. One can, however, still calculate RD [24].
We used chi-square test of contingency table to identify the difference of pooled incidence of ADRs between different dosage groups (≤ 10 mg/kg/day, 10-30 mg/kg/day, > 30 mg/kg/day). The number of studies in each category did not meet the criteria for conducting regression (at least 10 events per category of dependent variables and 10 events per category of independent variables).

Results
Of 10,700 titles identified, 445 proved potentially eligible after reviewing abstracts for the systematic reviews; 131 studies (133 articles) proved eligible following full text review. The updated search until September 2019 found two new trials (three articles) (Fig. 1). ESM Appendix 1 presents the characteristics of eligible studies. Risk of bias of individual RCTs was mainly due to unblinding of participants (high risk or unclear 65.1%, 56/86) or unblinding of outcome assessment (high risk or unclear 70.9%, 61/86). Risk of bias of individual cohort studies was mainly due to the lack of demonstration that outcome of interest was not present at the start of the study (55.9%, 19/34) and the lack of comparability of cohorts (76.5%, 26/34) (ESM Appendix 2). Among all types of studies, 4243 ADRs were reported from 197,675 pediatric patients who received azithromycin. The majority of ADRs (56.1%, 2382/4243) were reported from RCTs not as MDA, 14.2% (603/4261) from retrospective cohort studies, and 6.5% (274/4243) from prospective cohort studies (Table 1).

Azithromycin as MDA
We included eight RCTs and one prospective cohort study for azithromycin as MDA. Five RCTs reported no serious ADRs [25][26][27][28][29]. The incidence of ADRs was slightly higher in mass oral azithromycin communities compared with the untreated communities, and the most common ADRs were abdominal pain and vomiting in surveillance for adverse events during a large RCT [30]. The MORDOR study reported 11 cases of  (Tables 2 and 3). Compared with placebo, azithromycin showed no statistically significant difference in terms of incidence of ADRs (risk difference [RD] − 0.17 to 0.07; low to moderate certainty of evidence) (Fig. 2). Similar results were found when azithromycin was compared with cephalosporins and other macrolides (ESM Appendix 3). Azithromycin showed a decreased risk of diarrhea, when compared with amoxicillin/ clavulanate (RD − 0.11, 95% CI − 0.15 to − 0.07, P < 0.001; low certainty of evidence), but an increased risk of diarrhea when compared with penicillin V (RD 0.03, 95% CI 0.01 to 0.05, P = 0.005; high certainty of evidence) (ESM Appendix 3). ESM Appendix 4 presents certainty of evidence of each pooled estimate.

Retrospective studies
Three hundred and six ADRs were reported in 31,221 patients from eight retrospective cohort studies and 11 case report studies (ESM Appendix 1). The ADRs reported in retrospective studies were generally consistent with those in prospective studies. GI ADRs were still the most common events accounting for 44.9% (145/323) of all ADRs, followed by musculoskeletal and connective tissue ADRs (36.5%, 118/ 323). However, some ADRs that were not detected in prospective studies were reported by retrospective studies including tendon or joint disorders (TJDs) (n = 118), infantile IHPS (n = 8), and ventricular tachycardias (n = 4). The full list is shown in ESM Appendix 4.

Cardiac toxicity
Most prospective studies in children did not evaluate the risk of cardiac toxicity. Cardiac toxicity was only studied or reported in eight studies (six prospective, one retrospective, and one case report). Among prospective studies, five RCTs and one prospective cohort study reported 79 cardiac adverse events (Table 4) [35][36][37][38][39][40]. One prospective study where children received weekly azithromycin for 6 months reported statistically significant QT prolongation [35]. QT prolongation was also noted in two other studies. Two studies reported ADR, adverse drug reaction; MDA, mass drug distribution; RCT, randomized controlled trials *Nine RCTs and one prospective cohort study **One retrospective cohort study (n = 5039) focused on cardiac arrest in pediatric patients receiving azithromycin, and another retrospective cohort study (n = 15,073) focused on tendon or joint disorders We excluded 14 RCTs and 3 prospective cohorts from the calculation of pooled incidences of ADRs due to the lack of detailed description of ADRs ADR, adverse drug reaction; MDA, mass drug administrations *ADRs with pooled incidence less than 5 per 100 participants irregular heart rates and two studies reported no adverse events. A retrospective cohort study that compared azithromycin (n = 5039) with penicillin or cephalosporin (n = 77,943) in pediatric patients found that the rate of cardiac arrest in the azithromycin group was lower than that of the penicillin or cephalosporin groups (0.04% vs 0.14%, P = 0.04) [41]. A case report described severe bradyarrhythmia in a 9-month-old infant who received over 50 mg/kg azithromycin intravenously over 20 min [42].

Subgroup analysis of dosage
The incidence of diarrhea, vomiting, fever, and rash was higher in the high-dosage group compared with the low-or medium-dosage group (P < 0.01) ( Table 5). The incidence of prolonged QT and increased eosinophils was higher in the medium-dosage group than in the low-dosage group (P < 0.01) ( Table 5).

Discussion
Our review found that the main toxicity of azithromycin in pediatrics was gastrointestinal toxicity, specifically diarrhea, vomiting, and abdominal pain. Based on available data, the  [44,45]. For azithromycin not as MDA, our review highlighted that the most common toxicity was GI adverse reactions, and the most serious toxicities were cardiac adverse reactions and IHPS. The risk of cardiac toxicity and IHPS following MDA is unknown and can only be determined by prospective surveillance studies. The dose of azithromycin was associated with the risk of ADRs. The higher the dosage, the higher the risk of an ADR. Our study has several strengths. Using rigorous systematic review methods, we did a comprehensive search of the literature and evaluated the safety of azithromycin as both MDA and not as MDA. Our review included recently published studies which were not included in prior reviews, and thus, summarized all of the available evidence, providing optimal insight into the safety of azithromycin in pediatrics. We assessed the risk of bias of each primary study using risk of bias assessment tools based on study design. We detected the incidence of adverse events based on prospective studies from which the data are more reliable, and explored uncommon events from the larger retrospective studies.
This review also has some limitations. First, the ADRs were poorly reported in primary studies which led to the probable underestimate of the incidence of ADRs in this review. Even the meta-analysis might still be underpowered to detect potential differences between azithromycin and placebo or other antibiotics. The lack of ADR reporting is more common in studies of azithromycin as MDA and we were unable to conclude the safety of azithromycin as MDA compared with placebo. Second, we did not perform analysis of different age groups as planned in the protocol since most primary studies did not report the outcomes for each age group separately.
To our knowledge, the ADRs of azithromycin as MDA in children have not been evaluated in previous systematic reviews [46]. For azithromycin not as MDA, cardiac toxicity of azithromycin has been a concern for a long time. Previous studies and reviews found the risk of azithromycin appears to depend on age and prior cardiovascular risk in adults. However, none of them evaluated the risk in pediatric patients. In our review, cardiac adverse events were found in pediatric patients, but the difference between the azithromycin group and the control group was not significant in most of the studies. Only one study found a statistically significant increase in QT prolongation, but this study involved weekly administration of azithromycin for 6 months and only children who  received azithromycin received ECGs [35]. Additionally, the data from some studies must be questioned because the method for safety surveillance was poorly reported [26,27,35].
Future studies in pediatric patients should be conscious of using robust methods for safety monitoring. Previous studies have demonstrated the association between erythromycin exposure early in life and IHPS [47][48][49][50][51][52]. Only one retrospective study was found in our review that evaluated the association between azithromycin and IHPS. Since azithromycin and erythromycin are slightly different in molecular structure, further studies are required to determine the relationship between early exposure to azithromycin in life and IHPS [12]. The risk of IHPS following MDA is unknown and should be considered in future studies of MDA. However, prospective surveillance following MDA with larger numbers of patients is more likely to be beneficial.

Conclusion
The main ADRs of azithromycin whether used as MDA or not were gastrointestinal, specifically diarrhea, abdominal pain, and vomiting. For azithromycin not as MDA, the most serious toxicities were cardiac adverse reactions and IHPS. Increasing dose proved to increase the risk of ADRs.
Author contributions IC, LNZ, PPE, LLZ, TX, and SAQ designed the study. XCP, WYL, PPX, and XFN screened the literatures independently in two pairs. PPX, XCP, WYL, PPX, XFN, CC, and LSH abstracted data and cross-checked the data in pairs. LNZ and TX resolved discrepancies in literature screening and data abstraction. ZYB, LNZ, and PPX analyzed the data. LZN, PPX, and IC interpreted the data. LNZ, IC, TX, SAQ, DZM, and LLZ drafted the work or revised it critically for important intellectual content and have given final approval of the version published. All authors agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Compliance with ethical standards
Conflict of interest The authors declare that they have no conflict of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. There was no statistical difference between the two groups with the same letter a, b, or c in the following table. Otherwise, there is a statistical difference ADR, adverse drug reaction; RCT, randomized controlled trial *The total number of patients was 5811 in the low-dosage group, 2454 in the medium-dosage group, and 450 in the high-dosage group