Changing the Treatment Paradigm for Acute Otitis Media in Children

Too often, acute otitis media (AOM) is overdiagnosed, and too often, unnecessarily long antibiotic therapies are prescribed. Faced with the challenge of increasing resistance among bacterial pathogens causing AOM,^{1 - 2} an increased awareness that antibiotic treatment of AOM can provide important benefits to patients, but that spontaneous improvement also can occur,^{3 - 4} and calls to curtail and more judiciously use antibiotics for this common clinical condition⁵ signal it is time for change.

The quality of scientific data to support a policy for shorter courses of antibiotic treatment of AOM has been comprehensively reviewed.⁶ In this issue of THE JOURNAL, Kozyrskyj et al⁷ present a meta-analysis that provides a systematic, quantitative evaluation of the evidence. The 32 comparative trials analyzed were divided into 3 groups according to the pharmacokinetic behavior of the antibiotic used: (1) oral short-acting antibiotics, eg, amoxicillin (clavulanate), cefaclor, cefuroxime, cefprozil, and cefpodoxime proxetil (n=17); (2) oral azithromycin (n=11); and (3) intramuscular ceftriaxone (n=4). Among 1549 children who were treated for 5 days with short-acting antibiotics in comparison with 1569 children treated for 10 days, the difference in risk of failure was 7.8% favoring the longer treatment when the outcome evaluation was performed at 8 to 19 days after initiation of therapy. This difference was no longer evident at 1 month following treatment. Further, the marginal increased risk of treatment failure with 5 days of antibiotics was not documented when only the better-quality trials were included in the meta-analysis. No difference in outcome was observed in children treated with 3 or 5 days of oral azithromycin or with injectable ceftriaxone in comparison with those treated for 10 days with other antibiotics. Short-acting antibiotics given for less than 48 hours and to children with perforated eardrums did not appear to work well; failure occurred in 3 times as many children compared with treatment of 7 days or more. Sensitivity analyses were conducted to assess the robustness of the meta-analysis by reexamining the data after excluding trials of lower methodological quality, trials that included children with recurrent or chronic otitis media, and trials of comparisons between different antibiotics. The authors qualified their recommendation by acknowledging that subgroup sample sizes were too small to provide a reliable estimate of the risk of treatment failure with a shortened course of antibiotics in children younger than 2 years or in children with perforated eardrums. There was no evidence of publication bias. After reading this article, a series of questions came to mind.

First, how many children diagnosed as having AOM actually had the disease? The diagnosis of AOM can be difficult. Signs and symptoms of AOM are also observed in children without it. Several of the classic findings of AOM, such as fever and earache, are often absent even in cases confirmed by tympanocentesis. Most important, when a parent brings a child to the physician and reports sleeplessness, irritability, rhinorrhea, and perhaps fever, the temptation is great to see a little bit of redness or fluid behind the eardrum as justification for an antibiotic prescription.⁸ Thus, overdiagnosis may occur in 40% to 80% of patients in comparison with confirmation by tympanocentesis.^{9 - 10}

Neither the quality score of the trials included in the meta-analysis by Kozyrskyj et al⁷ nor the sensitivity analysis included a measure for accuracy in definition of disease, ie, the ability of the authors conducting the trials to make the diagnosis of AOM. This is a critical issue with respect to the external validity of the studies analyzed. Some of the trials included in the meta-analysis could have permitted inclusion of children who actually did not have AOM because tympanocentesis was not required for confirmation of the presence of middle ear fluid and pathogens. These children likely would be classified as cured or improved because they never had AOM, and an overestimation of efficacy for all treatments and a blurring of any actual differences in efficacy between regimens would follow.¹¹ This shortcoming is addressed in part by randomization of patients and by double-blind evaluations of treatment outcomes, criteria that were not always met in the studies included in the meta-analysis.

Second, will short-course antibiotic therapy cause more or less bacterial resistance? Most clinicians would agree that reducing the length of therapy minimizes exposure of bacteria to antibiotics and therefore reduces selective pressure for the emergence of resistance.¹¹ A recent study by Guillemot et al¹² suggests that longer duration of antibiotic treatment with an oral β-lactam antibiotic contributes to the selective pressure in promoting pharyngeal carriage of penicillin-resistant Streptococcus pneumoniae. Selection of bacteria with resistance to the antibiotic given and to multiple structurally unrelated antibiotics has been noted to follow more prolonged administration of these drugs.¹³ Antibiotics administered for shorter durations are therefore likely to produce less emergence of resistance.

Third, how many days do patients take their antibiotic? Poor compliance appears to be more common with longer antibiotic treatment courses because patients tend to stop taking their medications once symptoms have resolved.¹⁴ Typically, clinical resolution of AOM occurs within 2 to 5 days. Antibiotic courses that require fewer days are more acceptable to the patient and can enhance compliance.¹⁵

Fourth, is it biologically plausible that 5-day and 10-day antibiotic treatment courses are comparable in AOM? It appears that current treatment duration for AOM was derived from the success observed with 10 days of penicillin in the treatment of group A β-hemolytic streptococcal tonsillopharyngitis¹⁶ ; shorter regimens (5-8 days) are standard for AOM in other countries.¹⁷ Tympanocentesis obtained before and several days after onset of therapy suggest that susceptible bacteria usually will be eradicated by an effective antibiotic within 3 to 5 days after initiation of therapy.¹⁸

Fifth, can a 5-day course of antibiotics be used in children younger than 2 years with AOM? Children younger than 2 years represent about one third of the population that experiences AOM.¹⁹ In the meta-analysis by Kozyrskyj et al,⁷ a clear answer did not emerge regarding the duration of treatment in these younger children. Higher rates of treatment failure may occur in younger children compared with older children,¹⁴ and Paradise²⁰ has cautioned against 5-day therapy in children younger than 2 years.

Consideration of these issues and the results of the meta-analysis⁷ suggest that 2 changes should be made in the approach to management of AOM. First, the diagnostic accuracy in AOM needs to be improved. Pneumatic otoscopy has been advocated as an important adjunct to assist in diagnosis of AOM,^{21 - 24} yet most physicians find it inconvenient or remain unconvinced of its value. Adjunctive use of tympanometry and acoustic otoscopy can be of some assistance,²² but is more useful in follow-up than in acute diagnosis. The most accurate method to verify the diagnosis of AOM is tympanocentesis, which demonstrates the presence of middle ear fluid, drains the effusion, and allows subsequent culture for identification of causative pathogens.^{21 - 23} Tympanocentesis is not commonly performed by primary care physicians in their clinical practices, even in refractory cases of AOM. Instead, a sequence of different antibiotics is prescribed in the hope that one eventually will clear the presumed infection. In the absence of bacteriologic data provided by tympanocentesis, selection of second-line drugs is empiric and guided by the spectrum of activity of each drug.^{21 - 25} If the microbiologic cause of AOM can be documented on the basis of results of appropriate cultures of middle ear effusions obtained by tympanocentesis, then antibiotic selection can be more precise.^{21 - 25} A bacterial pathogen is generally isolated from middle ear fluid in two thirds of children with AOM²² and in 50% of children with persistent AOM²⁶ ; the balance require no antibiotic treatment.

More primary care physicians must learn and practice the use of tympanocentesis to improve diagnostic accuracy. Tympanocentesis should be performed if the patient appears toxic or is highly febrile or if the tympanic membrane is bulging to the point of anticipated spontaneous rupture.^{22 ,27} It is also of benefit in unresponsive AOM, which persists after a second course of antibiotics.^{21 ,23 ,27}

Second, the course of antibiotic treatment of AOM needs to be shortened. The data to support this approach have been comprehensively reviewed⁶ and now subjected to meta-analysis.⁷ Comparable efficacy between shorter- and longer-treatment courses is biologically plausible,¹⁸ and shorter courses of antibiotics may reflect more realistically the actual compliance behavior of patients.^{14 - 15} This strategy addresses issues of antibiotic overtreatment of AOM and concerns that such treatment contributes to the increasing problem of bacterial resistance.^{6 ,11} At a minimum, a 5-day treatment regimen as the standard would curtail antibiotic use for AOM by 50% for most patients. However, individualized therapy based on a physician's familiarity with the needs of a particular patient always should prevail. Ten or more days of antibiotic therapy may be necessary for children with tympanic membrane perforation and for children at higher risk for treatment failure, such as those with underlying medical conditions, those with chronic and recurrent otitis media, and those younger than 2 years. An important future research goal will be to identify subsets of children who are likely to benefit from longer antibiotic treatment of AOM.