Pediatrics · Medical Literature
The impact of topical or oral antibiotics in children with acute otitis media on their middle ear, nasopharyngeal, and gut microbiomes.
Abstract
Acute otitis media (AOM) is a major driver of paediatric antibiotic prescriptions. We assessed the impact of oral and topical antibiotics on middle ear, nasopharyngeal, and gut microbiome compositions, and the gut resistome, in children with AOM and ear discharge (AOMd). Fifty-eight children with AOMd and ear pain and/or fever were randomized to oral amoxicillin suspension (n = 31) or hydrocortisone-bacitracin-colistin eardrops (n = 27) for 7 days. From 57 out of 58 children, baseline, and Week-2 middle ear fluid (MEF) and nasopharyngeal (NP) samples were sequenced, along with baseline, Week-2, and Month-3 faecal samples. At baseline, the top 5 MEF genera were Streptococcus, Haemophilus, Turicella, Staphylococcus and Alloiococcus and NP genera Moraxella, Haemophilus, Streptococcus, Corynebacterium, and Dolosigranulum. At Week-2, the ear discharge had resolved in all but four children (oral n = 3, eardrops n = 1). In NP samples, the relative and absolute abundances of Streptococcus decreased to a greater extent after oral than eardrop treatment, but Moraxella and Haemophilus increased only following oral treatment. Neither treatment significantly altered the faecal microbiome or resistome at Week-2 and Month-3. Therefore, both treatments resolved the middle ear discharge in most children, but oral amoxicillin suspension may reduce NP Streptococcus more than hydrocortisone-bacitracin-colistin eardrops at the cost of potentially increasing other NP pathobionts.
Clinical relevance
In children with acute otitis media with ear discharge (AOMd), do oral amoxicillin and topical hydrocortisone-bacitracin-colistin eardrops differ in their short- and medium-term effects on middle ear fluid, nasopharyngeal, and gut microbiome composition, and on the gut antibiotic resistome? __SNAPSHOT__ {"studyDesign":"Randomized controlled trial (microbiome sub-study embedded within a previously reported open-label non-inferiority RCT); comparative two-arm parallel design with microbiome sequencing at multiple time points","setting":"52 primary care (general practice) practices in the Utrecht region, the Netherlands; December 2017 – March 2023","sampleSize":"n = 58 randomized (oral amoxicillin: n = 31; hydrocortisone-bacitracin-colistin eardrops: n = 27); 57 of 58 contributed analyzable samples","population":"Children aged 6 months to 12 years presenting with AOMd (acute otitis media with spontaneous tympanic membrane perforation and ear discharge) plus ear pain and/or fever; baseline characteristics reported as well-balanced between groups; median age 28 months in parent trial, 40% under 2 years","intervention":"Oral amoxicillin suspension 50 mg/kg/day in three divided doses for 7 days","comparator":"Topical hydrocortisone-bacitracin-colistin eardrops (5 drops, three times daily in the discharging ear[s]) for 7 days","primaryOutcome":"Not formally defined as a single primary microbiome endpoint in the abstract; study assessed changes in relative and absolute microbial abundances in middle ear fluid (MEF), nasopharyngeal (NP), and fecal samples, plus the gut antibiotic resistome, across three time points: baseline, Week 2, and Month 3","followUp":"3 months (microbiome sampling at baseline, Week 2, and Month 3 for fecal samples; baseline and Week 2 for MEF and NP samples)","fundingSource":"Not reported in the available abstract; parent trial was conducted in Dutch primary care with no commercial sponsor identified in prior publications"}
Key findings
At baseline, the five most abundant genera in middle ear fluid (MEF) were Streptococcus, Haemophilus, Turicella, Staphylococcus, and Alloiococcus; the five most abundant nasopharyngeal (NP) genera were Moraxella, Haemophilus, Streptococcus, Corynebacterium, and Dolosigranulum. Ear discharge resolved by Week 2 in 54 of 58 children (oral: 28/31; eardrops: 26/27), indicating high clinical resolution rates in both arms (calculated resolution rates: oral ~90%, eardrops ~96%). In NP samples at Week 2, relative and absolute Streptococcus abundance decreased to a greater extent following oral amoxicillin than eardrops; Moraxella and Haemophilus increased only in the oral amoxicillin group. Neither treatment significantly altered the MEF microbiome (abstract truncated before full MEF result reporting). Gut microbiome and resistome outcomes are referenced as assessed at baseline, Week 2, and Month 3, but specific effect sizes, confidence intervals, alpha- or beta-diversity statistics, and resistome gene abundance data are not reported in the available abstract. P-values, NNT/NNH, and between-group statistical comparisons for microbiome endpoints are not reported. Adverse event data specific to this microbiome sub-study are not reported in the available abstract.
Practice implications
This study established that oral amoxicillin produces measurable differential effects on the NP microbiome compared with topical eardrops, specifically greater suppression of Streptococcus and selective increases in Moraxella and Haemophilus — potential otopathogens — at Week 2. This pattern is biologically plausible given systemic versus local antibiotic exposure. The study did not establish that either treatment significantly altered the MEF microbiome (abstract truncation precludes full interpretation). Findings regarding gut microbiome composition and the resistome at Week 2 and Month 3 are referenced but not reported in detail in the available abstract, and cannot be clinically interpreted without full-text data. Only abstract-level data were available for this summary, which is a critical limitation. The findings apply to young children (predominantly under 3 years) with AOMd presenting to primary care; they may not generalize to children with AOM without perforation, immunocompromised children, children in high-prevalence antibiotic resistance settings, or non-European populations. The clinical importance of the observed NP microbiome shifts — particularly increased Moraxella and Haemophilus following oral amoxicillin — is uncertain and requires correlation with clinical outcomes such as recurrent AOM rates, which were not reported in the abstract. Practice should not change based on this microbiome sub-study alone; the parent non-inferiority RCT already demonstrated that eardrops did not meet the pre-specified non-inferiority margin for symptom resolution at Day 3. __LIMITATIONS__ Severely underpowered: only 58 of a planned 350 children were enrolled due to slow accrual over >5 years, leaving the study with very limited statistical power for all microbiome comparisons; risk of Type II error is high,Abstract-only data available: key quantitative outputs (diversity metrics, resistome gene counts, adjusted effect sizes, confidence intervals, p-values for all endpoints) are not reported in the provided abstract and cannot be evaluated,Open-label design: lack of blinding may introduce differential ascertainment or behavioral biases, though microbiome sequencing outcomes are objective,No placebo or untreated comparator arm: it is not possible to distinguish antibiotic-driven microbiome changes from disease-driven or spontaneous temporal changes,Abstract truncated: results for MEF microbiome, gut microbiome composition, and resistome outcomes are incompletely reported; the phrase 'Neither treatment significan—' was cut off, preventing full interpretation of the primary microbiome findings,Single-country, single-region primary care setting (Utrecht, Netherlands): findings may not generalize to populations with different baseline NP microbiome compositions, antibiotic resistance patterns, or vaccine coverage profiles,Sequencing methodology (16S rRNA amplicon vs. shotgun metagenomics), bioinformatic pipeline, and resistome detection platform are not specified in the abstract, limiting methodological assessment,Short follow-up for MEF and NP (Week 2 only): long-term NP microbiome recovery or sustained dysbiosis cannot be assessed,Treatment adherence differential (88% eardrops vs. 97% oral antibiotics in parent trial): lower adherence in the eardrops group may confound between-group microbiome comparisons,Funding source not reported: risk of sponsorship bias cannot be assessed __BOTTOM_LINE_CATEGORY__ promising_but_preliminary __BOTTOM_LINE__ In this severely underpowered microbiome sub-study of 58 children with AOMd, oral amoxicillin produced greater suppression of Streptococcus and selective enrichment of Moraxella and Haemophilus in the nasopharynx at Week 2 compared with topical antibiotic-corticosteroid eardrops, suggesting a systemic antibiotic-driven NP dysbiosis that topical treatment largely avoids. Gut microbiome and resistome data are referenced but not reported in the available abstract, precluding conclusions about the broader collateral effects of either treatment route. This preliminary, hypothesis-generating work supports further adequately powered trials incorporating microbiome endpoints to inform antibiotic stewardship for AOMd; no change in current prescribing practice is warranted based on this sub-study alone.