Participant flow and follow up—Of the 425 children with acute otitis media registered, 362 were eligible, and from these 240 were randomly assigned to one of the treatment groups (figure). Children in the antibiotic group and the placebo group differed in the prevalence of recurrent acute otitis media, regular attendance at a day care centre, and parental smoking habits (table ).
Baseline characteristics of 240 children randomised in trial of antibiotic use for treatment of acute otitis media. Figures are numbers of children except for mean age
Outcome at day four—Persistent symptoms at day four occurred in 69 out of 117 children (59%) in the amoxicillin group and in 89 of 123 (72%) in the placebo group (difference in risk 13%; 95% confidence interval 1% to 25%) (table ). Adjustment for recurrence, day care, and smoking as possible confounders in the logistic regression analyses showed an odds ratio of 1.79 (1.03 to 3.13). Among children with persistent symptoms four (one in the amoxicillin group and three in the placebo group) received other antibiotics. Three of these children were admitted to hospital (one in the amoxicillin group, two in the placebo group); one (placebo group) was admitted on the third day with meningitis but because of deterioration this child had already been started on another antibiotic on day two. The culture of cerebrospinal fluid yielded negative results, but the Gram stain suggested streptococcal meningitis. The two other children were admitted because of dyspnoea (amoxicillin group) and dehydration (placebo group). All four recovered without residual symptoms. Inclusion of the one child lost to follow up (amoxicillin group) in either outcome group did not materially change the findings.
Main outcome measures in infants with acute otitis media randomised to receive amoxicillin or placebo
Outcome at day 11—Clinical treatment failure at day 11 occurred in 72 out of 112 children (64%) in the amoxicillin group and in 84 of 120 (70%) in the placebo group (6%; −6% to 18%) (table ). Eleven children received other antibiotics (three in the amoxicillin group, eight in the placebo group) and were recorded as treatment failures. One of these children (placebo group) needed admission to the hospital because of deterioration of symptoms of acute otitis media. Six children (three in each group) were lost to follow up between day four and day 11, and in one case (amoxicillin group) the evaluation of the ear drum was missing, although the symptoms were gone. A best case scenario (amoxicillin group analysed as “cured” and placebo group as “not cured”) did not show any significant difference in clinical treatment failure at day 11 (9%; −3% to 21%).
Duration of fever and pain or crying—The median time to cessation of fever was two days with amoxicillin and three days with placebo (P=0.004; log rank test). Median time to cessation of pain or crying was eight days with amoxicillin and nine days with placebo (P=0.432; log rank test).
Analgesic consumption—During the first three days, mean analgesic consumption in the amoxicillin group was 1.7 doses and in the placebo group 2.5 doses (P=0.018). Over the whole 10 days these figures were 2.3 and 4.1, respectively (P=0.004).
Outcome at six weeks—At six weeks 212 children were examined. Middle ear effusion was present in 69/107 (64%) in the amoxicillin group and in 70/105 (67%) in the placebo group (3%; −10% to 16%). The proportion of children with bilateral effusion was 48% in both groups. In addition, no clear differences were observed between the two groups as regards recurrent acute otitis media, use of antibiotics in this period, referrals to the otolaryngologist or paediatrician, or surgery.
Adverse effects—De novo diarrhoea was reported on day four in 17% (20/117) of the amoxicillin group and in 10% (12/123) of the placebo group (difference −7%; −16% to 2%). On day 10 these figures were 12% (14/117) and 8% (10/123), respectively (difference −4%; −12% to 4%). Of the children lost to follow up, five were withdrawn (all between day four and day 11) because of possible side effects, two because of diarrhoea (both in the amoxicillin group) and three because of skin rashes (all in the placebo group).
Compliance—According to the diaries the mean number of doses of study medication taken was 24.6 (82% of possible total) in the amoxicillin group and 23.2 (76%) in the placebo group (P=0.9). According to the suspension remaining in the returned bottles, 80% of the children in both groups had received the full amount, and 95% received at least 80% of the amount prescribed.