Two rigorous controlled trials published in the New England Journal of Medicine indicate that in young children with a certain diagnosis of acute otitis media (AOM), a course of co-amoxiclav improved outcomes compared to placebo at the cost of increased adverse effects.

In the early days of antibiotics, AOM was regarded as a clear indication for antibiotic therapy, however trials carried out in the late 1970’s and 1980’s suggested that relatively few children gained significant benefit from such treatment. In view of the risks of adverse effects and bacterial resistance, guidelines began to recommend observation and symptomatic therapy as initial treatment. The reliability of the evidence behind these recommendations has been challenged, for example the use of less stringent diagnostic criteria in trials, inappropriate antibioitics, and exclusion of very young children, and the aim of both these trials was to examine the effects of an effective antibiotic in scrupulously diagnosed AOM in younger children.

The first study, in the US, included children aged 6 to 23 months with AOM diagnosed on strict criteria within 48 hours of inset. They were randomised to treatment with co-amoxiclav or matching placebo for ten days, stratified according to whether or not they had a history of recurrent AOM. Primary outcomes were the time to resolution of symptoms and the symptom burden over time; assessments were at 4 to 5 days, 10-12 days (end of therapy) and a post-therapy assessment at 21-25 days.

Of 1,385 children screened, 498 were eligible and 291 were enrolled into the trial (co-amoxicalv n=144, placebo n=147): main reasons for exclusion were not having AOM according to study criteria and lack of parental consent. For the primary outcome of time to symptom resolution, there was no significant difference in time to initial resolution: for co-amoxiclav, 35% had initial resolution of symptoms by day 2, 61% by day 4, and 80% by day 7; among children who received placebo, corresponding values were 28%, 54%, and 74% (P=0.14 for the overall comparison). For sustained resolution of symptoms, the difference was statistically significant: 20%, 41%, and 67% with co-amoxiclav, as compared with 14%, 36%, and 53% with placebo (P=0.04 for the overall comparison).

Children in the co-amoxiclav group were less likely to have clinical failure at either at 4–5 days (4% vs. 23%; difference, 19%; 95% CI, 12 to 27; P<0.001) and at or before the day 10–12 visit (16% vs. 51%; difference, 35%; 95% CI, 25 to 45; P<0.001).

More children in the co-amoxiclav group had adverse effects, most commonly diarrhoea (47.8% vs. 26.6%) or nappy-area rashes, and more were withdrawn due to adverse effects (7 vs. 2). One child in the placebo group developed mastoiditis.

The authors conclude that children aged 6 to 23 months with proven AOM benefit from treatment with co-amoxiclav, although they are more likely to have adverse effects. They comment that their placebo response was lower than that of previous trials, suggesting that this was due to the more stringent diagnostic criteria, and suggest that antibiotic treatment be restricted to those so diagnosed in order to reduce unnecessary antibiotic use.

The second study was carried out in Finland and involved children aged 6 to 35 months. They also had AOM diagnosed on strict criteria and were randomised to treatment with co-amoxiclav or placebo, in this study for 7 days. Primary outcome was a composite of time to treatment failure, consisting of six independent components: no improvement in overall condition by the first scheduled visit, a worsening of the child's overall condition at any time, no improvement in otoscopic signs by the end-of-treatment visit on day 8, perforation of the tympanic membrane at any time, severe infection necessitating systemic open-label antimicrobial treatment at any time, and any other reason for stopping the study drug (e.g., an adverse event or non-adherence to the study drug) at any time. Assessment was at day 3 and day 8 after enrolment.

Of 1,062 potential participants, 322 were randomised (main reasons for non-participation were not AOM according to study criteria or fell into exclusion criteria); one in each group received no medication and did not attend any study visits, and one (placebo) was withdrawn on day 1, leaving 319 for analysis. The primary outcome, treatment failure, occurred more often in the placebo group than in the co-amoxiclav group (44.9% vs. 18.6%; P<0.001). The difference was apparent by day 3 (25.3% vs. 13.7%) and the co-amoxiclav reduce the overall risk of treatment failure by 62% (hazard ratio, 0.38; 95% CI, 0.25 to 0.59; P<0.001). The estimated number to treat to avoid treatment failure in 1 child was 4 children (95% CI, 2.7 to 6.2).*

Adverse events were more likely in the co-amoxiclav group, occurring in 52.8% vs. 36.1% in the placebo group (an increase of 16.7 percentage points with co-amoxiclav; 95% CI, 5.8 to 27.6; P=0.003), diarrhoea being most frequent (47.8% vs. 26.6%; P<0.001). No patient developed mastoiditis.

The authors conclude that children aged 6 to 35 months with proven AOM benefit from effective antibiotic treatment, although they have more side effects. They suggest that further research should be carried out to identify those patients who would derive greatest benefit in order to minimise unnecessary antibiotic use.

An accompanying editorial discusses the two papers. The author points out that antibiotics became the standard of care in AOM because it was previously seen as a severe disease: he suggests that this may have been because only the worst cases were brought to medical attention, along with changes in diagnostic criteria, and in the virulence of the causative organisms. He commends the two studies for adding significant evidence showing that where children have certain AOM, they are likely to benefit from antibiotic treatment. He notes that many will recover without antibiotics, and that such treatment increases the risk of adverse effects; nevertheless we cannot tell at present which children will recover without antibiotics and which will not.

[* Editor’s note: 3.8 given in the paper – conventionally, NNT are rounded up to the next whole patient.]