We conducted a participant-blinded, care provider-blinded, assessor-blinded, randomised controlled trial — the Azithromycin versus Amoxycillin for Treatment of Acute Otitis Media in Aboriginal Children (AATAAC) study — comparing single-dose azithromycin with 7 days of twice-daily amoxycillin treatment. Treatment allocation was stratified by health centre, body weight and diagnosis (AOMwoP or AOMwiP), using randomly varying block sizes.

Our study commenced on 24 March 2003 and was completed on 20 July 2005. We screened Aboriginal children from 14 remote communities and two Aboriginal medical services in northern and central Australia. Our inclusion criteria were (a) age between 6 months and 6 years; (b) new diagnosis of AOMwoP or AOMwiP; and (c) willingness of parents to bring their child for a follow-up visit. Children were excluded if they had (a) previously been allocated to an intervention group in the same study; (b) received antibiotics in the previous 7 days; (c) current severe illness requiring intravenous or intramuscular antibiotic treatment to be given within the following 7 days; (d) a known allergy to penicillin or azithromycin; or (e) perforation covering more than 2% of the tympanic membrane.

Children with AOM were randomly allocated to receive either azithromycin (30 mg/kg as a single dose) and amoxycillin placebo, or amoxycillin (50 mg/kg/day in two divided doses for a minimum of 7 days) and azithromycin placebo. All children were scheduled to be examined on Day 0 and at the end of therapy (between Day 6 and Day 11). Children diagnosed with AOMwiP at the end of therapy received a second course of treatment and were also examined between Day 12 and Day 21.

All clinical assessments were made by ear health research officers. Otoscopic findings were recorded on a standardised form. Assessments were made using a GSI 38 tympanometer (Grason-Stadler, Eden Prairie, Minn, USA), a LumiView portable binocular microscope (Welch Allyn, Skaneateles Falls, NY, USA), a Siegle speculum for pneumatic otoscopy, and a video-otoscope (Welch Allyn, Skaneateles Falls, NY, USA). Video recordings of the tympanic membranes were reviewed by an independent unblinded observer.

We categorised middle ear states into one of six diagnostic categories, using criteria based on recommendations for clinical practice in this population:3

• Normal;

• Otitis media with effusion (intact and non-bulging tympanic membrane and type B tympanogram);

• AOMwoP (any bulging of the tympanic membrane and type B tympanogram);

• AOMwiP (middle ear discharge observed and perforation recently healed or present for less than 6 weeks or covering less than 2% of the pars tensa of the tympanic membrane [CSOM is nearly always associated with larger perforations]);

• Dry perforation (tympanic membrane perforation without any discharge observed); or

• CSOM (middle ear discharge observed and perforation present for longer than 6 weeks and covering at least 2% of the pars tensa of the tympanic membrane).

The final middle ear diagnosis reflected the child’s more severely affected ear (highest category).

The presence of nasal discharge was recorded if nasal discharge was visible from 1 metre away at any time during the clinical examination. Skin sores (raised crust or pus visible) were recorded if visible on the arms, legs or head.

Nasal swabs were taken at baseline and at follow-up examinations. Swabs of ear discharge were also collected on the days when nasal swabs were taken. All swabs were collected, transported, stored and cultured as previously described.14 Azithromycin resistance was defined as a minimum inhibitory concentration (MIC) of ≥ 2 µg/mL. High-level resistance was defined as MIC > 32 µg/mL.

We estimated that in at least 50% of children receiving standard therapy for AOM the condition would fail to resolve within 8 days. With 150 children in each intervention arm, the study would have a power of 95% to detect an improved outcome in an additional 20% of children (α = 0.05). This difference was chosen as the minimum improvement that would potentially result in a change in recommended therapy for AOM from amoxycillin to azithromycin.

All statistical analyses were conducted using Stata software, version 10 (StataCorp, College Station, Tex, USA). Assessment of statistical significance was based on the Fisher exact test.

Our study was approved by the human research ethics committees of the Menzies School of Health Research and the Royal Darwin Hospital and by the Central Australian Human Research Ethics Committee, and was registered with the Australian Clinical Trials Registry (ACTRN 12609000691246). An independent Data and Safety Monitoring Board chaired by Professor David Brewster and Associate Professor Alan Ruben of the NT Clinical School, Flinders University, reviewed the data.

Clinical observations. Of 320 children taking part in the trial, 165 were randomly allocated to the azithromycin treatment group and 155 to the amoxycillin treatment group. As seven children were lost to follow-up in each group, analysis was carried out on the remaining 306 children (Box 2). Most were less than 2 years of age, and about half were male. For each group at baseline, 85% had AOMwoP and 15% had AOMwiP, and 5% of mothers reported that they thought their child had ear pain. Nasal discharge was seen in 69% of children and skin sores in 6%–7%.

Nasal carriage. For both treatment groups, nasal carriage of S. pneumoniae and NCHi was 83% or more. About 14% of children were colonised by S. pneumoniae with intermediate or full resistance to penicillin (MIC ≥ 0.12 µg/mL), and about 5% carried azithromycin-resistant S. pneumoniae (MIC ≥ 2 µg/mL). Less than 10% of children swabbed carried β-lactamase-producing NCHi.

Ear discharge. In the azithromycin and amoxycillin groups, respectively, S. pneumoniae was cultured from 22% and 32% of ear discharge swabs, and NCHi from 36% and 41% of ear discharge swabs.

Clinical failure (Box 3). At the end of therapy, 50% of the azithromycin group and 54% of the amoxycillin group were clinical failures. Similar proportions (45% and 49%) failed to improve, and about 5% were worse in both groups (nine children developed new perforations). Differences between the azithromycin and amoxycillin groups in clinical failure rates were not significant.

No differences in clinical failure or failure to improve were indicated in a per-protocol analysis (children seen before Day 11 after commencement of treatment). Only four parents reported that their child had ear pain (three of the children were in the amoxycillin group). Fewer children in the azithromycin group than the amoxycillin group had runny nose (35% v 46%), but the difference was not significant, and about 4% of children in both groups had skin sores.

Clinical failure rates in children who had AOMwoP at baseline were 40% in the azithromycin group and 46% in the amoxycillin group. For children who had AOMwiP at baseline, clinical failure rates were 92% and 83%, respectively. (Clinical failure rates were 84% and 63%, respectively, at Day 12–21.)

There was no significant association between age group (< 2 years or ≥ 2 years) and clinical failure.

No significant difference in clinical failure was detected between azithromycin and amoxycillin groups for carriers or non-carriers of S. pneumoniae or NCHi at the end of therapy. Carriage of resistant S. pneumoniae or NCHi at baseline did not predict greater clinical failure. The risk difference (RD) in clinical failure between children carrying susceptible organisms (S. pneumoniae or NCHi) at baseline and those carrying resistant organisms at baseline was – 9%.

Complications and side effects. There were three serious adverse events that resulted in admission to hospital during the treatment period. The independent Data and Safety Monitoring Board found that these were not associated with the study interventions. No side effects (eg, allergic, gastrointestinal) led to cessation of treatment or withdrawal from the study.

New episodes of disease. In both groups, new perforations were detected in 3%–4% of children who had AOMwoP at baseline. Pain resolved in all children who had had pain at baseline, but four children developed “new” pain (Box 4).

Nasal carriage. Compared with children in the amoxycillin group, children in the azithromycin group had significantly lower carriage of S. pneumoniae and H. influenzae, lower carriage of S. pneumoniae with intermediate or full resistance to penicillin (RD, – 6% [but this difference was non-significant]), and significantly higher carriage of azithromycin-resistant S. pneumoniae (RD, 7%) (Box 5).

Ear discharge. The proportion of ear discharge cultures that were positive for S. pneumoniae and NCHi was about 25 percentage points lower in the azithromycin group (Box 5).

Compared with baseline, nasal carriage of S. pneumoniae was significantly lower in both groups at follow-up (azithromycin group RD, – 56% [95% CI, – 65% to – 47%]), and carriage of NCHi was also significantly reduced (azithromycin group RD, – 31% [95% CI, – 40% to – 21%]). In the azithromycin group, compared with baseline, carriage of S. pneumoniae with intermediate or full resistance to penicillin was lower at follow-up (10%) than at baseline (15%) (RD, – 5% [95% CI, – 12% to 2%]). In contrast, nasal carriage of azithromycin-resistant S. pneumoniae was higher at follow-up (10%) than at baseline (5%) (RD, 5% [95% CI, – 1% to 10%]).