Are antibiotics indicated as initial treatment for children with acute otitis media? A meta-analysisBMJ 1997; 314 doi: https://doi.org/10.1136/bmj.314.7093.1526 (Published 24 May 1997) Cite this as: BMJ 1997;314:1526
- Christopher B Del Mar, professor of general practicea,
- P Glasziou Paul, readera,
- Mauricio Hayem, postgraduate studenta
- a Centre for General Practice, University of Queensland Graduate School of Medicine, Brisbane, Australia 4006
- Correspondence to: Professor Del Mar
- Accepted 24 April 1997
Objective: To determine the effect of antibiotic treatment for acute otitis media in children.
Design: Systematic search of the medical literature to identify studies that used antibiotics in randomised controlled trials to treat acute otitis media. Studies were examined blind, and the results of those of satisfactory quality of methodology were pooled.
Subjects: Six studies of children aged 7 months to 15 years.
Main outcome measures: Pain, deafness, and other symptoms related to acute otitis media or antibiotic treatment.
Results: 60% of placebo treated children were pain free within 24 hours of presentation, and antibiotics did not influence this. However, at 2-7 days after presentation, by which time only 14% of children in control groups still had pain, early use of antibiotics reduced the risk of pain by 41% (95% confidence interval 14% to 60%). Antibiotics reduced contralateral acute otitis media by 43% (9% to 64%). They seemed to have no influence on subsequent attacks of otitis media or deafness at one month, although there was a trend for improvement of deafness at three months. Antibiotics were associated with a near doubling of the risk of vomiting, diarrhoea, or rashes (odds ratio 1.97 (1.19 to 3.25)).
Conclusions: Early use of antibiotics provides only modest benefit for acute otitis media: to prevent one child from experiencing pain by 2-7 days after presentation, 17 children must be treated with antibiotics early.
There is wide variation in the use of antibiotics for early treatment of acute otitis media in children, and we examined the literature by meta-analysis to establish what benefits or harm antibiotics provide
Antibiotics did not influence resolution of pain within 24 hours of presentation, though at 2-7 days after presentation, by which time only 14% of children in control groups still had pain, early use of antibiotics reduced the risk of pain by about 40%
Antibiotics also reduced contralateral acute otitis media but seemed to have little influence on subsequent attacks of otitis media or deafness
Antibiotics were associated with a near doubling of the risk of vomiting, diarrhoea, or rashes
Early use of antibiotics provides only modest benefit for acute otitis media: to prevent one child from experiencing pain by 2-7 days after presentation, 17 children must be treated with antibiotics early
Acute otitis media is extremely common in children. By the age of 3 months, 10% of children will have suffered at least one episode. The incidence peaks between 6 and 15 months.1 In Western countries mortality is low, but it may be higher in underdeveloped countries.2 Complications are now rare in the West, although in 1954 the rate of mastoiditis was 17% in cases of acute otitis media.3 Symptoms consist mainly of pain and systemic illness, sometimes very distressing, which in 80% of children is limited to 24 hours' duration.4 The pain is caused by pressure on the tympanic membrane, which can sometimes be seen bulging and inflamed at otoscopic inspection. After the inflammation settles, the consequent deafness left by fluid retained within the middle ear space may take several weeks to resolve.
Considerable attention has been focused on the role of infection in acute otitis media. Several attempts at identifying causative infectious agents have yielded several, the main ones being Streptococcus, Branhamella catarrhalis and Haemophilus species.5 Some viruses have been implicated. Twelve different case series failed to identify a causative infectious agent in the middle ear fluid of 28-62% of patients.5 The details of the pathophysiological process, traditionally described as arising from the increased bacterial load and obstructive elements that occur during an upper respiratory tract infection, may be incompletely understood.
There is wide variation in the use of antibiotics between the doctors of different nations, from as low as 31% of cases of acute otitis media in the Netherlands to as high as 98% in Australia and the United States.6 We examined the literature by meta-analysis to establish what benefits or harm antibiotics provide for patients with acute otitis media. Because acute otitis media is a disease that remits spontaneously the notion of “cure” is not meaningful. We examined the health outcomes of resolution of symptoms (pain and deafness) and the most commonly reported serious complication (mastoiditis) without reference to signs. While other end points (such as microbiological “cure”) may enhance an understanding of the disease process, we regarded them as only of secondary interest in this empirical study of effectiveness.
We manually searched titles in Index Medicus from 1958 to 1965 and, by computer, searched Medline and Current Contents from 1966 to August 1994 using combinations of “OTITIS MEDIA” and a search strategy previously described for optimally identifying controlled trials.7 The references of all retrieved studies were searched as well. All identified randomised controlled trials of antimicrobial drugs versus placebo control were included. The data we extracted consisted of severity and duration of pain (midterm and long term), deafness, adverse effects, and recurrent attacks.
In assessing the quality of the methodology of each study identified, we adapted a protocol described previously to attribute scores8: for the manner in which subjects were assigned to treatment or control group; control of selection bias after assignment to treatment (trials analysed on an intention to treat basis were preferred, and where necessary and possible intention to treat analyses were reconstructed); adequacy of blinding; and objectiveness of assessment of the outcome. Scores could range from 0 (worst possible) to 11 (best possible). The method used is available from us. By cutting and pasting, we assessed the studies blind to the authors, institutions, journal, and results of each study. The three of us met to resolve differences in our independent assessments still blind to the identity of each study.
We performed χ2 tests for heterogeneity of the odds ratio for all analyses. These showed no significant heterogeneity. We used the Peto method to calculate combined estimates for a fixed effects model for the odds ratio and performed a z test of significance. All calculations were done with REVMAN 2.0 (Update Software, 1995).
Eight trials were eligible for inclusion in our review of antibiotics against placebo. One had a factorial design (treatment by myringotomy, antibiotics, both, or neither), of which we used only the antibiotic and placebo arms.9 One study did not report on empirically relevant, patient centred outcomes.10 Another reported only recurrences.11 Thus, only six studies of children aged 7 months to 15 years were available for analysis. Studies allowed for children in the trials who were not doing well to be removed and treated with antibiotic after the code was broken. This occurred at different rates (14%,12 8%,13 and 7%14).
The methodological quality of the six selected studies was good (see table 1). Five used a blinded randomisation and outcome assessment. Two failed to include all children in follow up assessments, although data were missing for less than 10%.
Figure 1) shows the outcomes of the studies. About 60% of placebo treated children were pain free within 24 hours of presentation, and antibiotic treatments did not influence this. However, at two to seven days after presentation, antibiotics reduced pain in the remaining children by 41% (95% confidence interval 14% to 60%). They similarly reduced the risk of developing contralateral acute otitis media by 43% (9% to 64%), and they showed trends for reducing perforations of the tympanic membrane. Antibiotics seemed to have no influence on subsequent attacks of otitis media or deafness at one month (as estimated from tympanometry), although there was a trend for a benefit at three months. They were associated with a near doubling of the risk of problems commonly associated with antibiotics including vomiting, diarrhoea, and rashes (odds ratio 1.97 (1.19 to 3.25)).
The number of well conducted studies is small for such a common condition.17 As all were conducted in Western countries, the results may not be generaliseable to Third World communities, where the far greater risk of serious suppurative complications may support the early use of antibiotics.2 There seems to be a deficiency of research on this subject.
Implications of analysis
Several matters must be considered when deciding the implications of our findings. Initial use of antibiotics will reduce pain, and contralateral otitis media, by a relative reduction of about 40%. However, there is benefit only for those 14% of patients whose pain has not spontaneously resolved within 24 hours of presentation. This is equivalent to an absolute benefit of 5.6% fewer children experiencing pain by two to seven days after presentation. Thus, 17 children must be treated at first presentation to prevent one child experiencing pain after two to seven days, which is of the same order as a previous meta-analysis of the subject.4 Many children suffering contralateral otitis media will be counted among those with persistent ear pain. It is not surprising antibiotics provide no pain relief within the first 24 hours when you consider the steps required for obtaining, ingesting, and absorbing antibiotics and for starting antibiotic activity.
Looking for subgroups of children with otitis media who would benefit from antibiotics might be a useful aspect of research. Knowing which children are going to suffer an illness extending beyond one day would enable doctors to select and treat only those who would benefit. Although we found some evidence of prolonged symptoms with placebo treatment among young children, those with previous episodes of otitis media, and those with bilateral acute otitis media, the differences were small.12
Antibiotics seem to have little effect on deafness, particularly deafness that is not prolonged. This is surprising in view of a recent report that antibiotics may assist in managing glue ear.18
Implications of not using antibiotics
What are the likely consequences of not using antibiotics? For 17 months, 60 general practitioners in the Netherlands used nose drops and analgesia alone for initial treatment of acute otitis media in all children aged 2-12 years. Only 3% (136/4860) of these children suffered a severe course of the illness (that is, child still ill after 3-4 days or ear discharge for more than 14 days).19 This proportion is far smaller than the results for the control groups in this meta-analysis would suggest. Two of the children developed mastoiditis, but this settled uneventfully after treatment with amoxycillin.19 Subsequent follow up of these general practitioners indicated that most still seldom used antibiotics to treat otitis media and that mastoiditis remained rare.20
Many doctors and their patients may be disinclined to use antibiotics at first presentation of otitis media for so little benefit. Others may regard any potential benefit as worth the inconvenience of purchasing and administering the drugs and the risk of their (usually) minor complications. Perhaps the best approach is to regard antibiotics as an optional treatment for early acute otitis media, together with adequate analgesia, that doctors should discuss openly with their patients. In future, studying what influences doctors' decisions whether to use antibiotics might be more fruitful than undertaking more trials of the treatment itself.
Conflict of interest: None.