Executive Summary

Publication Details

Introduction

Acute Otitis Media (AOM)1 is a viral and/or bacterial infection of the middle ear and represents the most common childhood infection for which antibiotics are prescribed in the United States. Timely and accurate diagnosis and management of AOM can have significant individual and public health consequences.

The 2001 AHRQ evidence report on the management of AOM analyzed the evidence on the initial management of uncomplicated AOM in children, focusing on the natural history of the disease and the use of antibiotics in management. Although the 2001 report provided valuable analysis of the literature on the management of uncomplicated AOM in children, it did not address issues related to diagnostic accuracy and precision, management of AOM in specific subgroups of children, or the impact of immunization with Heptavalent Pneumococcal Conjugate Vaccine (PCV7) on the microbiology of AOM, recommended for widespread use in 2000. Additionally, new trials of treatment continue to be published. The purpose of this current AHRQ evidence report is to examine and analyze the evidence on three broad areas of inquiry: 1) accuracy and consistency of the clinical diagnosis of AOM, 2) the impact of PCV7 on AOM microbial epidemiology, and 3) the comparative effectiveness of different treatment options for uncomplicated AOM in average risk children and in children with recurrent (defined as three or more episodes in six months or four or more episodes within 12 months) or persistent AOM.

Methods

Key Questions

The American Academy of Pediatrics, the nominating organization, proposed six key questions aimed at assessing the comparative efficacy of interventions to treat uncomplicated and recurrent AOM in terms of treatment success, the safety of such treatments, and the effect on children in specific subgroups. In conjunction with a technical expert panel we refined these questions:

  1. Diagnosis of AOM: What are the operating characteristics (sensitivity, specificity, and likelihood ratios) of clinical symptoms and otoscopic findings (such as bulging tympanic membrane), both individual and composite, to diagnose uncomplicated AOM and to distinguish it from otitis media with effusion (OME)?2
  2. What has been the impact of the Pneumococcal Heptavalent Immunization (PCV7) on AOM microbial epidemiology (including acute mastoiditis and suppurative complications), with respect to both the organisms associated with AOM and the patterns of antimicrobial resistance?
  3. What is the comparative effectiveness of different treatment options for treating uncomplicated AOM in average risk children?
  4. What is the comparative effectiveness of different management options for recurrent otitis media (uncomplicated) and persistent otitis media or relapse of AOM?
  5. Do treatment outcomes in Key Question3 (KQ3) and KQ4 differ by characteristics of the condition (AOM), patient, environment, and/or health care delivery system, including but not limited to the following: A. Laterality, i.e., unilateral vs. bilateral; B. Otorrhea or perforation; C. AOM severity, i.e., as defined as defined by the AAFP/AAP AOM Guideline (2004); D. Comorbidities, e.g., asthma; E. Age groups, e.g., <4 weeks, 4 weeks to <6 months, 6mos–<2 years, 2–5 years; F. Race; G. Ethnicity; H. Day care attendance?
  6. What adverse effects have been observed for the treatments whose outcomes are addressed in KQ III and KQ IV?

Literature Searches

Searches of PubMed and the Cochrane Databases of Systematic Reviews, Cochrane Central Register of Controlled Trials, and Education Resources Information Center were conducted from January 1998 through July 2010 using the same search strategies used for the 2001 report, with the addition of terms for conditions not considered in the 2001 review (recurrent otitis media), new drugs, and the heptavalent vaccine. The Web of Science was also used to search for citations of the 2001 report and its peer-reviewed publications. Among the 8, 945 titles identified were a number of recent, good-quality systematic reviews, which were included and which were examined for references. Titles were screened independently by two pediatricians with experience in conducting systematic reviews. For the question pertaining to diagnosis, we searched primarily for studies that included an assessment of sensitivity and specificity relative to a defined gold standard; we identified one good-quality 2003 meta-analysis and replicated its search strategy to obtain subsequent studies not included in their analysis. For the question pertaining to the effect of the vaccine on epidemiology and microbiology, we searched for studies that compared microbiology in the same populations before and after introduction of the vaccine or studies that compared microbiology across vaccinated and unvaccinated populations. For the efficacy and safety questions, we searched primarily for controlled trials or large observational studies aimed at identifying adverse effects.

Literature Review, Data Abstraction, and Analysis

In total, the reviewers examined 8,945 titles for the draft version of this report; 739 titles were identified for further review. Of those, 72 articles that met the predetermined inclusion criteria were reviewed in detail for efficacy and safety results. Investigators abstracted data into standard evidence tables with abstraction checked by a second investigator. Studies were quality-rated by two investigators using established criteria. For randomized controlled trials (RCT), the Jadad criteria were used. QUADAS criteria were used to evaluate the studies that pertained to diagnosis. Data abstracted included parameters necessary to define study groups, inclusion/exclusion criteria, influencing factors, and outcome measures. Data for the analysis were abstracted by a biostatistician and checked by a physician reviewer. We used a sequential resolution strategy to match and resolve the screening and review results of the two reviewers.

For the assessment of treatment efficacy, pooled analysis was performed for comparisons for which three or more trials could be identified. The articles eligible for analysis for the key questions pertaining to treatment efficacy were grouped according to the specific treatment options they compared. Each comparison consisted of articles that were considered homogeneous from the standpoint of clinical practice. Since the question of treatment efficacy was addressed in the first evidence report published in 2001, we combined the articles identified in that report with articles newly identified for this evidence report that addressed the same populations and reported the same types of outcomes. We pooled data for comparisons that included three or more articles from the old and new searches and performed meta-analyses or quantitative syntheses. We used the Der Simonian and Laird random effects model to pool rate differences across studies. Among the three effect measures—rate difference, relative risk, and odds ratio—the Technical Expert Panel and the project staff chose as most suitable the rate difference and its 95 percent confidence interval. We also reported the findings on the success rate instead of the failure rate throughout the report as recommended by the Technical Expert Panel. A test of heterogeneity was performed using the I2 statistic. GRADE criteria were applied to assess the quality of the evidence for each comparison. In addition to the pooled estimate, we report the Q statistic and p-value for the Chi-squared test of heterogeneity.

For the assessment of the adequacy of evidence in arriving at a conclusion on the effectiveness of a particular treatment using a particular outcome, we use the concept of the “minimal clinically important difference (MCID)” against which the location of the 95% confidence interval of the pooled outcome was compared. Confidence intervals falling within the zone of MCID were considered to establish evidence of no difference, and confidence intervals outside the zone of MCID were considered to establish difference. If the confidence intervals crossed into the zone of MCID, an effect (positive or negative) of the treatment option on the outcome could not be established. While the MCID for treatment of AOM has not been empirically determined, we used an MCID of 5%, as this value represents approximately the lower limit of what Cohen would classify as a “small” effect size for treatment of AOM. Users of this evidence report who consider larger or smaller differences to be the minimum clinically important effect may reach different conclusions than we do here.

Results

Key Question I. Diagnosis of AOM: What Are the Operating Characteristics (Sensitivity, Specificity, and Likelihood Ratios) of Clinical Symptoms and Otoscopic Findings (Such As Bulging Tympanic Membrane) to Diagnose Uncomplicated AOM and to Distinguish It from OME?

Three clinical criteria are necessary to diagnose AOM: 1. acute symptoms of infection, 2. evidence of acute tympanic membrane (TM) inflammation, and 3. presence of middle ear effusion (MEE). To address this key question, we searched for studies that examined clinicians’ accuracy and precision in identifying each of these clinical criteria, or their accuracy and precision in identifying all three together. A 2003 systematic review and three additional original studies met the inclusion criteria for the present review. The systematic review found that among symptoms, only otalgia (ear pain) (sensitivities of 54%, 60%, 100% in three different studies; specificities 82%, 92%; positive likelihood ratio [LR] 3.0 [2.1–4.3], 7.3 [4.4–12.1]) and ear rubbing (sensitivity 42%; specificity 87%; positive LR 3.3 [2.1–5.1] seemed to predict a clinical diagnosis of AOM. An article published subsequent to the 2003 review found that among 469 children ages 6–36 months with parent-suspected AOM in primary care offices, AOM diagnosis was not associated with the occurrence, duration, or severity of parent-reported symptoms (e.g., ear pain: sensitivity 92%, specificity 8%, positive LR 1.0 [1.0–1.1]; ear rubbing: sensitivity 70%, specificity 22%, positive LR 0.9 [0.8–1.0]; fever: sensitivity 43%, specificity 65%, positive LR 1.2 [1.0–1.6]).1

One of the studies examined in this 2003 review assessed the accuracy of individual physical exam findings (cloudy, bulging, immobile, or red TM); they found these signs to be positively associated with AOM determined by the presence of MEE on tympanocentesis and clinical symptoms.

A study published subsequent to the 2003 review examined the accuracy of otoscopic and tympanometric findings compared with tympanocentesis as the criterion standard to determine the presence of MEE. The investigators performing otoscopy were not blinded to the tympanogram (a tool that evaluates middle ear function) results; further, the criterion standard of tympanocentesis was performed only when otoscopic or tympanometric findings suggested MEE. Ninety-seven percent of children with MEE on tympanocentesis had “Type B” tympanogram findings (abnormal), and all children with MEE on tympanocentesis had an otoscopic exam consistent with AOM. However, positive LR estimates are not as useful, since all participants had an AOM diagnosis at enrollment.

The second study published subsequent to the review included 137 eardrums that were either assumed to be or were diagnosed as AOM by general practitioners (GP). Of these, 78% were confirmed by ear-nose-and-throat (ENT) exam and the remaining were not, because the otolaryngologist diagnosed OME, viral otitis, or a normal TM. The ENT exam confirmed the GP diagnoses more often when redness and bulging were noted by the GP (83%) than when redness only was noted (75%).

The prior review and three additional studies that we identifieded for this key question did not directly or completely answer it; however, the studies do suggest that clinical findings of MEE (decreased mobility or abnormal position) and middle ear inflammation (distinctly red color of the TM) are positively associated with AOM, defined by positive tympanocentesis and acute onset of symptoms. Further, studies comparing diagnostic accuracy between generalist or primary care physicians and otolaryngologist suggest that clinicians’ accuracy in identifying all three clinical criteria in one patient is moderate, at best. The overall quality of evidence for this Key Question is considered low, meaning that further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.

Key Question II. What Has Been the Impact of the Pneumococcal Heptavalent Immunization (PCV7) on AOM Microbial Epidemiology: What Organisms (bacterial and viral) are Associated with AOM Since the Introduction of PCV7; and What Are the Patterns of Antimicrobial Resistance in AOM Since the Introduction of PCV7?

Two types of studies could address this question: observational studies that compared the types of organisms associated with AOM among children prior to and following introduction of the PCV7 vaccine in 2000 and RCTs of vaccine efficacy that compared the causative agents between a group of unvaccinated children and those who were vaccinated. Both study types are complementary. RCTs provide a better assessment of cause-and-effect for the relationship between the vaccine and changes in organisms, but often enroll highly restricted patient populations. Observational studies complement RCTs by providing data on more representative populations.

We identified six original studies (four observational studies and two RCTs) that provided some information on this question. Since the introduction of PCV7, the observational studies generally report that Haemophilus influenzae (HF) has become more prevalent as a causative agent of AOM and Streptococcus pneumoniae (SP) has become less prevalent, although SP remains an important agent as well. The introduction of the vaccine has also resulted in a greater proportion of non-vaccine serotypes and a smaller proportion of the vaccine serotypes. The RCTs provided findings consistent with those results.

We were also asked to assess the evidence for subpopulations of children according to prior antibiotic use. However we found no studies that analyzed the effects of the vaccine on causative agents according to whether the children had or had not received antibiotics in the past.

The overall quality of evidence for this Key Question is considered high for the conclusion that use of the PCV7 vaccine has resulted in shifts in the prevalence of causative agents, meaning further research is very unlikely to change our confidence in the estimate of effect. The quality of evidence is very low for the special populations (such as patients with recurrent or persistent AOM) since we found fewer studies examining the vaccine’s effect on these special populations.

Key Question III. What Is the Comparative Effectiveness of Different Treatment Options for Treating Uncomplicated AOM in Average Risk Children?

For the comparison of treatment success for children with uncomplicated AOM, we identified 63 comparisons of treatment options for uncomplicated AOM that encompassed different antibiotics and regimens. Our analyses yielded inconclusive results for many of these comparisons. For 12 comparisons, we reached stronger conclusions. Table S-1 shows key comparisons from the first AOM report, the present report, and where possible, combined results.

Table S-1. Comparative Effectiveness of Different Treatment Options for Treating Uncomplicated Acute Otitis Media (AOM) in Average Risk Children in the 2001 Report and the Present Report.

Table S-1

Comparative Effectiveness of Different Treatment Options for Treating Uncomplicated Acute Otitis Media (AOM) in Average Risk Children in the 2001 Report and the Present Report.

Meta-analyses of the comparison of ampicillin or amoxicillin vs. placebo indicates that nine children (95% CI: 6, 20) with uncomplicated AOM would need to be treated with immediate antibiotic therapy rather than placebo to note a difference in the rate of clinical success by day 14. For the comparison of ampicillin or amoxicillin vs. placebo, the quality of evidence is moderate due to heterogeneity in the results of studies, with the higher quality studies reporting smaller benefits, meaning that further research is likely to have an important impact on our confidence in the estimate of the effect and may change the estimate. In four studies of delayed treatment approaches for uncomplicated AOM, (1) two had higher rates of clinical success with immediate antibiotic therapy, i.e. Little (2001) and McCormick (2005) individually demonstrated higher clinical success rates for amoxicillin than for prescription-to-hold at day 3 (NNT=6; 95% CI: 4, 17) and wait-and-see at day 12 (NNT=7; 95% CI: 4, 17) options, respectively, (2) two did not demonstrate a difference in clinical success between immediate vs. delayed antibiotics, and (3) three studies showed a marked decrease in antibiotic utilization in the delayed antibiotic group.

Four trials, one newly identified for this report and three identified for the original AOM report addressed the comparison of ampicillin or amoxicillin vs. ceftriaxone. No difference (RD=0%, 95% CI: −7, 7) was found between these treatments for clinical success by day 14 though this finding was inconclusive utilizing an MCID of 5% (one trial found a slight advantage for ceftriaxone, whereas the others found ceftriaxone to be slightly less effective). The quality of evidence for this conclusion is moderate, meaning that further research is likely to have an important impact on our confidence in the estimate of the effect and may change the estimate.

Five trials, two newly identified and three identified for the original AOM report, compared amoxicillin-clavulanate (7–10 days) with single-dose ceftriaxone. No difference (RD=3%, 95% CI: −2, 7) was found between these treatments for clinical success by day 16 though this finding was inconclusive utilizing an MCID of 5%. The quality of evidence for this conclusion is moderate, meaning that further research is likely to have an important impact on our confidence in the estimate of the effect and may change the estimate.

Meta-analysis of three studies demonstrated equivalence of day-14 clinical success rates (RD=−0.7%, 95% CI: −4, 3) between cefaclor (7–10 days) and azithromycin (≤ 5 days) in treatment of uncomplicated AOM. In addition, single studies of comparisons (that could not be pooled) produced strong results. The quality of evidence for this conclusion is considered high, meaning further research is very unlikely to change our confidence in the estimate of effect.

In pooled analysis, no difference (RD=−0.3%, 95% CI: −7, 6) was noted in clinical success at day 14 comparing amoxicillin-clavulanate to azithromycin though this finding was inconclusive utilizing an MCID of 5%. In a single study, amoxicillin-clavulanate (for 10 days) was shown to have higher clinical success rates than azithromycin (single dose, one day) by day 14 when the pathogen was HF (NNT=4, 95% CI: 2, 17) and higher success rates than cefaclor by day 34 when success was defined by clinical symptoms (NNT=4, 95% CI: 2, 17). The quality of evidence for this conclusion is moderate due to heterogeneity in the results of studies, meaning that further research is likely to have an important impact on our confidence in the estimate of the effect and may change the estimate.

Equivalent clinical success rates were demonstrated in individual studies of amoxicillin vs. azithromycin, amoxicillin vs. erythromycin, amoxicillin-clavulanate vs. amoxicillin-sulbactam, cefixime vs. ampicillin or amoxicillin, cefaclor 50 mg/kg/day vs. 40 mg/kg/day, and amoxicillin-clavulanate 45/64/mg/kg/day divided into two daily doses vs. 40/10/mg/kg/day divided into three daily doses. In addition, individual studies of amoxicillin-clavulanate >60mg/kg/d vs. amoxicillin-clavulanate 40mg/kg/d and high-dose amoxicillin bid vs. lower-dose amoxicillin tid that in the 2001 Report were assessed as demonstrating equivalent clinical success rates are now assessed as inconclusive utilizing an MCID of 5%. Each of these single study results requires replication before strong conclusions can be reached.

Key Question IV. What Is the Comparative Effectiveness of Different Management Options for Recurrent Otitis Media (Uncomplicated) and Persistent Otitis Media or Relapse of AOM?

In approaching this question, studies were divided into those that examined treatment and those that examined prevention.

The available evidence did not allow us to reach strong conclusions regarding the following comparisons identified by this study for treatment of AOM in children with ROM, persistent AOM, or AOM treatment failure: amoxicillin-clavulanate vs. gatifloxacin, amoxicillin-clavulanate vs. levofloxacin, and amoxicillin-clavulanate vs. azithromycin. The overall quality of evidence for these comparisons is considered low, meaning that further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. One systematic review and additional new studies were identified examining otic antibiotics for treatment of AOM in children with tympanostomy tubes; however, it was not clear from the reports if the tympanostomy tubes were placed for ROM, persistent AOM, or some other chronic middle-ear condition, so these results cannot be generalized.

Several prior systematic reviews addressed the prevention of AOM in children with ROM. One review concluded that long-term antibiotics, defined as six weeks or longer, decreased episodes of AOM from 3 to 1.5 (95% CI: 1.2, 2.1) for every 12 months of treatment per otitis-prone child during active treatment. However data are missing regarding the safety of long-term antibiotic administration and the potential consequences on bacterial resistance. The role of tympanostomy tube placement was examined in a pooled analysis of two studies. This analysis found that tympanostomy tubes played a significant role in maintaining a disease-free state in the first six months after tube insertion in children with ROM. This conclusion is qualified by the small number of studies included in the analysis.

The available evidence did not allow for any definitive conclusions about the comparative role of amoxicillin vs. azithromycin, amoxicillin vs. sulfisoxazole, amoxicillin vs. placebo, sulfisoxazole vs. placebo, ceftibuten five-day vs. 10-day, probiotics vs. placebo, sulfafurazole vs. adenoidectomy, adenoidectomy vs. placebo, adenoidectomy vs. adenotonsillectomy, adenotonsillectomy vs. placebo, and adenoidectomy plus tympanostomy vs. tympanostomy in preventing AOM in children with ROM. The overall quality of evidence for each of these comparisons is considered low, meaning that further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.

Key Question V. Do Treatment Outcomes in Key Question3 (KQ3) and KQ4 Differ by Characteristics of the Condition (AOM), Patient, Environment, and/or Health Care Delivery System?

Of the 48 randomized clinical trials newly identified in our review that addressed the effectiveness of treatment options in uncomplicated AOM, 15 trials reported analyses for subgroups stratified by age, presence of MEE, laterality, parent/caretaker, hearing deficit presence/severity, otorrhea, examiner, and pneumococcal vaccine. Of the 10 trials identified in our review that addressed the effectiveness of treatment options in ROM, three reported analysis by age subgroups, and one reported stratified analysis by laterality and severity of otitis media.

For uncomplicated AOM, the available evidence indicated that antibiotic effect may be modified by age, laterality, and otorrhea. Definitive conclusions could not be made regarding subgroup analyses by other characteristics of AOM such as severity, characteristics of the patient such as presence of hearing deficit, characteristics of the environment such as the primary daytime caretaker, or characteristics of the healthcare delivery system such as the examiner.

In general, the results of individual trials and of meta-analyses show that children over the age of 2 have better outcomes from AOM, regardless of whether they are treated with antibiotics or not, compared to children 2 years of age or younger. No differences were seen in our meta-analyses in the rate difference for treatment success between children younger or older than 2 years when comparing ampicillin/amoxicillin to placebo or when comparing amoxicillin clavulanate to azithromycin. Similar conclusions were found in an individual patient meta-analysis.

In general, the results of individual trials and meta-analyses show that children with bilateral disease responded as well to treatment as those with unilateral disease. If left untreated, children with unilateral disease did better than those with bilateral disease. Further, the effect of antibiotic (compared with placebo) was greater in children with otorrhea than in those without otorrhea.

Key Question VI. What Adverse Effects Have Been Observed for the Treatments Whose Outcomes Are Addressed in KQ3 and KQ4?

We examined the incidence of adverse events in the RCTs identified for this report that compared the effectiveness of one or more treatment options. We also searched the FDA MedWatch Database for adverse events associated with use of medications for the treatment of AOM; however, none could be identified.

In general we could not make definitive conclusions regarding differences in adverse event rates among antibiotics when taking into account a MCID of 5%. However, Table S-2 shows the significant differences in adverse event rates that we noted (Table S-2 also shows the comparisons for the original report, those unique to the present report, and those that could be combined across both reports). Adverse events were generally more frequent for amoxicillin-clavulanate than for cefdinir, ceftriaxone, or azithromycin.

Table S-2. Comparison of Rates of Adverse Events Between Drugs (Significant Differences Only).

Table S-2

Comparison of Rates of Adverse Events Between Drugs (Significant Differences Only).

Of the 44 RCTs newly identified for this report that compared the effectiveness of treatment options in uncomplicated AOM, there are 61 treatment comparisons. Of the 61 treatment comparisons, 42 included comparisons of the percent of cases that had experienced an adverse event between two treatment options. For treatment of uncomplicated AOM, five adverse event rate comparisons showed a significant difference between two treatment options. Amoxicillin-clavulanate was associated with diarrhea more often than was cefdinir (NNT=four) and more often than was ceftriaxone (NNT= seven). The adverse event rates ranged from 27% to 35% for amoxicillin-clavulanate and from 10% to 14% for the other treatment options. For mention of any adverse event, amoxicillin-clavulanate had a higher rate than cefdinir given once or twice daily and a higher rate than ceftriaxone. However, in one study, the dose of amoxicillin was 40mg/kg/day, whereas in the other study, it was 80mg/kg/day (the clavulanate dosage was 10mg/kg/day in both studies). Equivalence was demonstrated in 29 comparisons, leaving 99 comparisons inconclusive.

These findings complement the findings from the first review, which showed that for uncomplicated AOM, children treated with amoxicillin-clavulanate for seven to ten days had a 19% (95% CI: 9, 29; NNT=5) higher rate of overall adverse effects and a 18% (95% CI: 8, 28; NNT=6) higher rate of gastrointestinal adverse effects than children treated with five days of azithromycin. (Although it was not specified in the studies, the original formulation was 31.25 mg clavulanate per 125 mg of amoxicillin). Eight children would need to be treated with azithromycin rather than amoxicillin-clavulanate to avoid a gastrointestinal adverse event. The original review also found that children treated with cefixime had an 8% (95% CI: 4, 13; NNT=12) greater rate of diarrhea than children treated with ampicillin or amoxicillin, so 12 children would need to be treated with ampicillin or amoxicillin rather than cefixime to avoid one case of diarrhea.

We also examined adverse event rates in children with presumed or explicitly defined ROM who were being given antibiotics for the treatment or prevention of AOM. Among the fourteen studies focused on children with ROM, persistent AOM, or AOM treatment failure, there were 21 treatment comparisons: eight involving the treatment of AOM in children with presumed or explicitly defined recurrent and/or persistent AOM, and/or AOM with treatment failure and the remainder in children being given the drugs prophylactically for prevention of AOM. For treatment of AOM in children with ROM and/or persistent otitis media, and/or AOM with treatment failure, we found one study that identified a significant difference in adverse event rates. In that study, amoxicillin-clavulanate (amoxicillin 90mg/kg/day; clavulanate 6.4mg/kg/day) was associated with diarrhea more often than was ciprofloxacin-dexamethasone ear drops (NNT=5). However, in 41 other comparisons, the adverse event rates were equivalent. In 23 comparisons, a definitive conclusion was not possible. For studies that examined prevention of AOM in children with ROM, we did not find any significant differences in any of the adverse event rate comparisons.

Conclusions

This section begins with a brief review of the limitations identified for this review. We then present our conclusions and recommendations for future research.

Limitations

The conclusions that can be drawn from this review of the evidence are limited by a number of factors, some associated with specific questions and some that cross the entire body of literature.

  • Assessing the precision of methods used to diagnose AOM is severely limited by the continued absence of a true gold standard and the reliance on the clinical definition. Although tympanocentesis is employed as the gold standard in some studies, its reliability and validity are limited by the need for specially trained operators, and studies that use tympanocentesis rarely perform the procedure on asymptomatic ears.
  • Assessing the possible impact of the PCV7 vaccine on AOM microbial epidemiology and the development of antibiotic resistance is limited by several factors. First, tympanocentesis is not routinely done in children with uncomplicated AOM. Thus, most of the studies that compared the microbiology of AOM before and after the introduction and use of PCV7 examined middle-ear fluid samples for children with complicated, recurrent, or persistent OM. Another limitation is that we do not have adequate data to understand the possible impact of PCV7 on non-bacterial agents (i.e., viruses). Although the importance of non-bacterial agents has been studied for AOM, we were unable to find studies examining the impact of PCV7 on the importance of non-bacterial causes of AOM.
  • The assessment of treatment efficacy was limited by the finding that the definitions of clinical success were usually not equivalent among studies comparing the same treatments. For example, studies used different clinical criteria to define success, and success was often measured at different time points. Another limitation to our assessment of treatment efficacy is that because we pooled studies across different time periods, we could not take temporal changes in microbiology into account, that is older studies might have had a microbiology more (or less) responsive to antibiotics than newer studies.
  • The inclusion criteria for participantss also varied widely among studies. Some studies used only one of the three criteria included in the definition of AOM for diagnosis, while others considered two or all three. It is possible that some studies with less stringent inclusion criteria may have included participants who did not have AOM, but rather had OME or no middle ear infective process at all. In addition, if the operating characteristics of criteria used to diagnose AOM differ by age, then it is possible that treatment outcomes by age may be confounded by a differential rate of inclusion of children who actually do not have AOM into a particular age group.
  • Few studies assessed the effect of patient characteristics on treatment outcomes, beyond the effect of age, laterality, or otorrhea.
  • Studies that compared adverse effects between treatments almost never explicitly included the collection of adverse event information in their designs and were rarely, if ever, powered to assess differences in rates of adverse effects between treatments. In addition, differences in the ways adverse events were reported and categorized from one study to another made it difficult to try to pool these results.

Discussion

AOM is a clinical diagnosis with three components: acute signs of infection and evidence of middle ear inflammation and effusion. 12 Evidence suggests that certain otoscopic findings (i.e., a red and immobile or bulging TM) predict AOM, but the accuracy or precision of a clinical diagnosis has not been determined. Given the absence of a gold standard for diagnosing AOM, it is difficult to draw firm conclusions from existing studies or to design new studies to assess the precision of diagnostic methods or criteria for diagnosing AOM. Perhaps the most important way to improve diagnosis is to increase clinicians‘ ability to recognize and rely on key otoscopic findings. Since the introduction of the PCV7 vaccine, AOM microbiology has shifted considerably. Our review indicates that overall, the SP serotype is becoming less prevalent, yet still important, while HF is increasing in its importance as an infectious agent of AOM. No studies that fit the inclusion criteria for the report examined the impact of the introduction of PCV7 on antimicrobial resistance.

For the treatment of uncomplicated AOM, immediate ampicillin/amoxicillin treatment has a modest benefit compared to placebo or delayed antibiotics, but also may be associated with more diarrhea and rash. Of 100 average-risk children with AOM, we could expect approximately 80 to get better within about 10 days without antibiotics. If all were treated with immediate ampicillin/amoxicillin, we would expect an additional 12 to improve, but 3 to 10 children would develop rash and 5 to 10 would develop diarrhea. Clinicians need to weigh these risks (including possible long-term effects on antibiotic resistance) and benefits before prescribing immediate antibiotics for uncomplicated AOM.

In head-to-head comparisons, most antibiotic regimens demonstrated comparable clinical success rates. Because of the relatively small number of studies on treatment of AOM in children with ROM, we are unable to draw any definitive conclusions regarding the comparative effectiveness of different antibiotic treatments. The evidence suggests that long term antibiotics decrease episodes of AOM from three to 1.5 for every 12 months of treatment per otitis-prone child during active treatment. However, the drawbacks of long-term antibiotics, which include adverse effects such as diarrhea, allergic reactions, and emergence of bacterial resistance, must be weighed against that of recurrence. Further, we can also conclude that tympanostomy tubes can help decrease the likelihood of a repeat infection in a child with a history of ROM within the first six months after tube insertion. This conclusion may be tempered by the issue of AOM diagnostic accuracy in the presence of tympanostomy tubes possibly confounding these results, i.e. the pressure equalization and drainage afforded by the tubes and their physical presence decreasing the intensity or visibility of signs and symptoms used to diagnose AOM, leading to false negatives. Again, whether or not the benefit of avoiding a repeat episode of AOM over six months outweighs the costs of a tympanostomy tube placement will depend on the clinician‘s assessment of the child with AOM, and discussions of advantages and disadvantages with the family.

While the 2001 evidence review identified only sufficient evidence to allow the assessment of the effects of age on treatment effectiveness, the current review identified information to assess the effect of laterality and otorrhea as well. The current review suggests that overall, children over the age of two years had better outcomes with various antibiotic options than children under age two and that laterality and otorrhea do have effects as well. These findings suggest that clinicians may need to more closely monitor response to treatment and outcomes when treating very young children with AOM, in particular those with bilateral AOM and those with otorrhea.

Although the evidence was generally insufficient to allow definitive conclusions regarding differences in adverse event rates, the available evidence across all studies did indicate an increased rate of gastrointestinal effects and diarrhea specifically with amoxicillin-clavulanate (compared with oral cefdinir, oral ceftriaxone, or ciprofloxacin-dexamethasone ear drops) and with cefixime (compared with ampicillin or amoxicillin). In addition amoxicillin-clavulanate appeared to have a higher overall adverse effect rate than cefdinir, ceftriaxone, or azithromycin.

Future Research Suggestions

Based on the findings of this review, we provide the following suggestions for future research directions.

Diagnosis of AOM

Additional studies are needed to more fully understand the precision of the current diagnostic criteria for AOM: acute onset of signs and symptoms, MEE, and middle ear inflammation. For example, although it has been determined that all three are necessary for a diagnosis of AOM, evidence is insufficient to guide clinicians on the most effective and efficient ways to assess each of these elements in the clinical setting. Also needed are more studies that use a reference standard that can take into account all three criteria of an AOM diagnosis. Thus, a reference standard that takes into account only MEE does not provide sufficient evidence on overall diagnostic accuracy for AOM.

Influence of the PCV7 Vaccine on Microbiology/Epidemiology

Studies are needed to address the implications of the observed evolution in microbiology subsequent to introduction of the PCV7 vaccine. For example, will this shift in microbiology translate to a shift in the type and incidence of suppurative and other complications? Further research is needed to explore the impact of PCV7 on the clinical progression and outcomes of uncomplicated AOM, and of AOM in otitis-prone children with recurrent AOM.

More inquiry is needed into microbiologic shifts in AOM, especially as it relates to resistance patterns of the non-PCV7 serotypes of SP that seem to be increasing since the introduction of PCV7. Such research will require continued surveillance of both shifts in the causative organisms of AOM and in the antibiotic resistance/susceptibility of these organisms.

A recent study of a single pediatric practice, not meeting our inclusion criteria, found evidence suggesting that an increase in the proportion of AOM with non-vaccine SP serotypes may be leading to another shift in AOM microbiology.8 These new data support the need for ongoing surveillance of AOM isolates.

Continued surveillance will also help us understand the impact of new pneumococcal vaccines that include more serotypes than PCV7 currently does, such as the newly-licensed PCV13. It will be important to have information to help conduct cost-benefit analysis of vaccines that cover more than the current seven serotypes. A growing body of research is assessing the efficacy of the vaccine in preventing AOM. Although a review of this literature was beyond the scope of this report, such a review may be warranted in the near future.

Treatment Efficacy and Adverse Effects

Research issues identified in the original AOM review are still applicable to the review update as it relates to treatment of uncomplicated AOM as well as to treatment of ROM, which was not previously addressed. Though we report several definitive conclusions, the usefulness of these conclusions to the practitioner is limited because of concerns regarding the internal validity of some of the source studies and the generalizability of the findings because of differences in the definitions of AOM and ROM—as well as treatment outcomes—across studies; the variability of study quality; and the relative paucity of evidence related to influencing factors such as characteristics of AOM including severity, the patient, the environment, and the healthcare delivery system. Standard definitions of AOM and ROM that lead to standard diagnostic criteria and that are acceptable to both researchers and practitioners have not been developed since the initial review and are still needed. The continued diversity of definitions for AOM as well as for ROM and, therefore, the diversity of diagnostic criteria that control entry of participants into these treatment trials make it difficult to synthesize and generalize findings, as it is unclear if the same condition is being assessed across studies. Greater knowledge regarding the effect of children’s age on the operating characteristics of diagnostic criteria will also help to assess results of studies comparing treatment options, e.g., by clarifying whether children of different ages who have been diagnosed with and are being treated for AOM truly have the condition. In addition, improved knowledge of the effect of tympanostomy tube presence on these diagnostic operating characteristics will help to better assess the true impact of tympanostomy tubes on prevention of AOM in children with ROM.

Standard definitions related to the quality of AOM management in terms of specific structures, processes, and outcomes are still needed. Differences in terminology and in particular outcome choice and definitions between studies make it difficult to synthesize the results across studies and to generalize findings. This issue should be addressed in future studies.

Footnotes

1

A diagnosis of AOM requires 1) a history of acute onset of signs and symptoms, 2) the presence of middle ear effusion (MEE), and 3) signs and symptoms of middle-ear inflammation. (Marcy, Takata, Shekelle, et al., 2001).

2

Otitis media with effusion (OME) is defined as fluid in the middle ear without signs or symptoms of acute infection. Distinguishing AOM from OME often poses a diagnostic challenge.