- Jack Froom, professor of family medicinea,
- Larry Culpepper, professor of family medicineb,
- Max Jacobs, associate professor of family medicinec,
- Ruut A DeMelker, professor of family medicinec,
- Larry A Green, Woodward-Chisholm chairman of family medicined,
- Louk van Buchem, otorhinolaryngologiste,
- Paul Grob, professor of general practicef,
- Timothy Heeren, associate professor of biostatistics and epidemiologyb
- a State University of New York, Stony Brook, NY 11794, USA
- b Boston University, Boston, MA 02118, USA
- c University of Utrecht, 3584 CG Utrecht, Netherlands
- d University of Colorado, Denver, CO 80220, USA
- e St Elizabeth Hospital, 500 LC Tilburg, Netherlands
- f University of Surrey, Guildford, Surrey GU2 5YG
- Correspondence to: Professor Larry Culpepper Department of Family Medicine, Dowling 5 South, 1 Boston Medical Center Place, Boston, MA 02118-2393, USA.
- Accepted 28 May 1997
Increasing worldwide resistance of bacteria to antimicrobial drugs is causing a crisis, manifested by higher morbidity, mortality, and costs.1 In 1992 the Institute of Medicine in the United States warned of the growing threat posed by resistant bacteria,2 and in 1994 the Centers for Disease Control in Atlanta initiated a prevention strategy,3 linked to a global plan by the World Health Organisation.4 Proposed remedies include development of new antimicrobials, improved sanitation, and educating patients not to ask for antimicrobials when they are not useful and physicians to prescribe them conservatively.5
About 30% of British children under the age of 3 visit their general practitioner for acute otitis media each year6 and 97% receive antimicrobials.7 In America it is the most common reason for outpatient antimicrobial use. Because this use has uncertain benefits, it merits reconsideration.
Increasing resistance to antimicrobial agents has been reported for the three most common bacterial causes of otitis media (Streptococcus pneumoniae, Haemophilus influenzae, and Moraxella catarrhalis), but rates differ between countries.8 9 In England and Wales in 1990-5, resistance to penicillin by S pneumoniae increased from 1.5% to 3.9% and to erythromycin from 2.8% to 8.6%.10
Except in the Netherlands, antimicrobials are standard treatment for acute otitis media in most developed countries.7 11 Although type and duration vary between countries, generally ampicillin, amoxycillin, or co-amoxiclav (amoxycillin-clavulanate) are preferred, with co-trimoxazole a low cost alternative.7 In the Netherlands, treatment of symptoms without antimicrobials has been adopted as routine initial treatment for otitis media,12 and this policy is associated with decreased emergence of resistance among organisms commonly found in otitis media.13 14 This approach also is being adopted in Iceland.15
Does treatment improve outcomes?
Seven randomised blinded studies have compared antimicrobials with placebo in patients with acute otitis media (table 1)).16 17 18 19 20 21 22 23 Comparisons are difficult because inclusion criteria, outcome measures, duration of follow up, and the period when studies were conducted are different. Also some studies had insufficient power to evaluate small to moderate differences. Nevertheless, since results are mixed and no study found large differences between placebo and antimicrobial groups, we conclude that the benefit of routine antimicrobial use for otitis media, judged by either short or long term outcomes, is unproved.
Bacterial resistance to antimicrobials is responsible for increasing morbidity, mortality, and costs
The most frequent use of antimicrobials in the United States is for otitis media
Evidence from randomised, placebo controlled trials that routine use of antimicrobials decreases the duration and severity of symptoms and prevents complications is weak
Treatment of acute otitis media differs worldwide, and careful use of antimicrobials in the Netherlands and Iceland seems to have reduced rates of resistance among organisms without compromising outcomes
The management of acute otitis media needs reassessing
A recent meta-analysis concluded that one child out of seven benefits from treatment,24 but only four of the 33 studies included had placebo or no medication groups,16 19 21 19 and no justification was given for excluding several methodologically sound placebo controlled studies.18 20 22 In contrast, after reviewing 50 trials, Claessen et al judged that methodological flaws precluded making recommendations.26
Does treatment prevent complications?
Although preventing mastoiditis and meningitis is a rationale for antimicrobial treatment, little evidence exists that routine treatment is effective for this purpose. Mastoiditis, a commonly reported complication of otitis media in the preantimicrobial era, is now uncommon. It is uncertain, however, whether the decrease is related to antimicrobial treatment, changes in clinical course and organism virulence, or increased host resistance. Also, uncomplicated otitis media often might not have come to medical attention in the era before antimicrobials, increasing the relative rate of reported complications.
Questions related to the routine use of antimicrobials for acute otitis media
Are the course of and outcomes after acute otitis media improved by antimicrobial treatment?
Do antimicrobials prevent complications of acute otitis media?
Which children are at high risk of poor outcomes after acute otitis media? Do they benefit from antimicrobials?
If antimicrobials are used, are the optimum type and duration of administration known?
Have management options, other than routine antimicrobial administration, been widely used and with what results?
Does decreased use of antimicrobials for acute otitis media influence the development of antimicrobial resistance?
Selective use of antimicrobials might yield the same benefit as routine use. In the Netherlands, among 4860 consecutive patients with acute otitis media who were not given antimicrobials, only two experienced mastoiditis; both responded to treatment with oral antimicrobial as outpatients.27 Justification for routine antimicrobial treatment of otitis media to prevent meningitis is equally weak. None of the 4860 patients developed meningitis. Kilpi et al reported that positive blood cultures in children with bacterial meningitis were equally common in those treated and not treated with antimicrobials before admission (77% and 78%).28 Even when otitis media and meningitis coexist, they are likely to result from a common upper respiratory infection with haematogenous spread, rather than direct spread from ear to meninges.29 It therefore remains uncertain whether either routine or selective antimicrobial treatment prevents the complications of mastoiditis and meningitis.
Do children at high risk benefit from antimicrobials?
Adverse consequences of otitis media include lack of bacteriological or clinical resolution, recurrent infection, persistent effusion, hearing loss, adenoidectomy, and insertion of a tympanostomy tube. No study has addressed whether antimicrobial treatment decreases the frequency of these in all or some of those with known risk factors.
The most important risk factors for poor outcome are young age and attendance at day care centres. In a study of two groups of children infected with S pneumoniae, comparable by sex, age at first attack, and frequency of earlier attacks, more penicillin resistant strains were isolated in those under 18 months (P=0.003).30 Decreased rates of bacteriological or clinical resolution7 22 31 and increased rates of recurrence,32 adenoidectomy, and insertion of a tympanostomy tube33 are other consequences reported in children under 2 years old.
Day care is a risk factor for poor outcomes, including recurrent infection,34 admission to hospital, adenoidectomy, and insertion of a tympanostomy tube.33 After other risk factors (age <2 years, white race, male sex, and history of tonsillitis, enlarged adenoids, or asthma) were controlled for, children in day care had a 50% higher chance of repeated ear infections than those not in day care.35 In a nine country study, children in day care were more likely to have a history of poor hearing, tympanostomy tubes, tonsillectomy, or adenoidectomy and to be referred to an otolaryngologist at the initial visit.36
Other factors likely to contribute to poor recovery include multiple previous episodes,37 bottle feeding,38 history of ear infections in parents or siblings,39 and use of a dummy.40 Although exposure to tobacco smoke has been considered a contributor to poor outcome, the evidence is controversial and the effect weak at best.41 Although risk factors are known, no study has shown that antimicrobial treatment improves outcome in children at risk.
What is the optimal type and duration of administration?
In the nine country study, antimicrobials did not improve outcome at two months, and no differences in rates of recovery were found for either antimicrobial type or duration.7 The optimum duration of antimicrobial treatment for otitis media is uncertain. There is evidence that two days,42 three days,43 five days,44 and 10 days are equally effective. Treatment for more than a few days might have little local effect. In a study of antimicrobial penetrance of the middle ear, penicillin concentrations decreased by about 70% after the second day of treatment (compared with the first day), suggesting that penetration depends on inflammation.45
No compelling evidence
After addressing these four questions, we conclude that existing research offers no compelling evidence that children with acute otitis media routinely given antimicrobials have a shorter duration of symptoms, fewer recurrences, or better long term outcomes than those who do not receive them. It also is not clear that routine compared with selective use of antimicrobials prevents complications. Thus it is prudent to reconsider routine use of antimicrobials for otitis media and to consider other approaches.
What are the effects of other management options?
Dutch family physicians use antimicrobials for upper respiratory tract infections more conservatively than doctors in other countries (see box).18 27 A World Health Organisation study of tonsillitis in 17 European countries found that 68% of Dutch patients received antimicrobials compared with 94% in other countries.46 In 1990 the Dutch College of General Practitioners adopted a guideline for treating acute otitis media. It has received widespread use in the Netherlands, but its adoption elsewhere requires consideration of the current healthcare system, particularly the ability of parents to access care for their children initially and if they fail to improve after treatment of symptoms.
Patients 2 years and older
Treatment of symptoms only (paracetamol with or without decongestant nose drops) for the first three days
Re-evaluation if symptoms (pain or fever thought to be due to acute otitis media) continue for three days. At that time the doctor may continue additional observation or give an antimicrobial (amoxycillin, or erythromycin if amoxycillin is contraindicated) for seven days
Special treatment for tympanic membrane perforation is not suggested unless it persists for 14 days, at which time a course of antimicrobials is suggested
Children between the ages of 6 months and 2 years
Management is the same as for those 2 years and older, except for a mandatory contact (either telephone or visit) after 24 hours. If there is no improvement doctors may either start antimicrobials or wait an additional 24 hours
Referral to an otolaryngologist suggested if patients in this age group appear to be seriously ill or do not improve after 24 hours of treatment with antimicrobials
Has this guideline resulted in more complications in Dutch children? While no controlled study has addressed this question, van Buchem et al found that only 2.7% of 4860 children given symptomatic treatment and no antimicrobial drug developed persistent fever, pain, or persistent discharge after three to four days and only two developed mastoiditis.27 Compared with cases in seven other countries where antimicrobial therapy is virtually universal, Dutch patients had similar outcomes at two months.7 Thus Dutch experience suggests feasible alternatives to antimicrobial treatment exist.
What is the effect on antimicrobial resistance?
Antimicrobial use in children with otitis media results in the emergence of resistant organisms in those children and in the community. Among children previously given antimicrobials for otitis media, Harrison et al found a rate of ampicillin resistance in bacteria in effusions obtained during a subsequent bout of otitis media that was three times higher than normal.47 Significant increases in resistant strains of H influenzae, M catarrhalis,48 and S pneumoniae 49 have been reported among children with otitis media who have had previous antimicrobial treatment. Development and spread of multiply resistant pneumococci after treatment for otitis media have been documented in day care centres and surrounding communities,50 including instances that have led to deaths from meningitis in children treated previously for uncomplicated acute otitis media.51
Although the organisms that cause otitis media are similar across countries,52 the Netherlands has a lower prevalence of resistant strains than other European countries. A study in 1989 in the Netherlands found that only 3% of 1100 S pneumoniae cultures showed decreased sensitivity to penicillin compared with more than 10% in most countries.14 53 Similarly, 4.7% of non-type B H influenzae isolates from respiratory tract infections were resistant to amoxycillin—less than the 10% overall rate in Europe.13
In Iceland, penicillin resistant pneumococci appeared in 1988 and increased to nearly 20% of pneumococcal infections by 1993.54 After a campaign to decrease antimicrobial use, particularly for otitis media, and to control spread of infections in day care centres, this rate declined to 16.9% in 1994. The presence of resistant pneumococci among carriers who are day care attendees declined from 20% (21 of 104) in 1992 to 15% (19 of 126) in 1995.15 Nevertheless, it is not possible to state with certainty that decreased use of antimicrobials results in decreased resistance in endemic bacteria.
What needs to be done?
Research—Placebo studies indicate that more than 80% of children with acute otitis media recover without antimicrobials. Antimicrobials might be useful for some of the remainder, but evidence of benefit is lacking. A trial should be designed to study antimicrobial effectiveness in children at high risk for poor outcomes, using carefully chosen outcome measures, with sufficient numbers of patients for subgroup analyses.
Prevention of acute otitis media—Sufficient information exists to support aggressive encouragement of breast feeding and avoidance of tobacco smoke.38 55 There is insufficient evidence of effectiveness to recommend pneumococcal vaccine.56 Over 70% of cases of acute otitis media are immediately preceded by viral respiratory infections,31 and there is convincing evidence that contaminated hands and surfaces are more important than airborne respiratory droplets in viral transmission.57 58 Campaigns to encourage day care centres to increase hand washing by staff and children and the use of virocidal agents to wipe surfaces might reduce viral transmission and subsequent otitis media.
Treatment of acute otitis media—Clinicians should immediately reconsider the routine use of antimicrobials for children with otitis media and consider treating symptoms with analgesics and observation for lack of improvement. Children less than 2 years old require frequent re-evaluation if not treated. Doctors are often uncertain about the diagnosis of otitis media 7; in such circumstances in particular antimicrobials should be withheld. If antimicrobials are used they should be given for less than 10 days. In addition doctors should identify modifiable risks for poor outcome and work with parents to eliminate them. The potential benefits of these actions are profound. They could reduce costs without compromising outcomes while reducing the risk of bacterial resistance.
This paper is dedicated to the memory of Dr Max Jacobs, whose intellect, wit, and ability to solve problems internationally made our collaboration possible. He died suddenly on 21 December 1996.
Funding: Agency for Health Care Policy and Research, Grant No. RO1 HS07035-03.
Conflict of interest: None.