The log-rank test was used for comparisons. A, Time to normal middle ear status was defined as having normal tympanometry findings (A curve) from both ears on 2 consecutive days (primary outcome) (P = .02). B, Time to improved middle ear status was defined as receiving a peaked tympanogram finding (A or C curve) from both ears on 2 consecutive days (P = .001). C, Time to normal pneumatic otoscopy finding (P = .002).
Tapiainen T, Kujala T, Renko M, Koivunen P, Kontiokari T, Kristo A, Pokka T, Alho O, Uhari M. Effect of Antimicrobial Treatment of Acute Otitis Media on the Daily Disappearance of Middle Ear EffusionA Placebo-Controlled Trial. JAMA Pediatr. 2014;168(7):635-641. doi:10.1001/jamapediatrics.2013.5311
Antimicrobial treatment reduces the symptoms of acute otitis media (AOM). The effect of antimicrobial treatment on the duration of middle ear effusion (MEE) and concomitant hearing impairment is not known.
To determine whether the antimicrobial treatment of AOM reduces the duration of MEE.
Design, Setting, and Participants
This randomized, double-blind, placebo-controlled trial involved a total of 84 children with AOM between 6 months and 15 years of age. Participants were recruited from September 14, 1999, to January 4, 2000; October 10, 2005, to December 16, 2005; and September 22, 2009, to June 4, 2012, from among children attending an AOM prevention trial and children visiting local outpatient clinics in Oulu, Finland.
Children were randomly allocated to receive either 40 mg/kg of amoxicillin-clavulanate or a placebo mixture per day for 7 days.
Main Outcomes and Measures
The primary outcome measure was the time to the disappearance of MEE as defined by a normal tympanogram finding (A curve) from both ears on 2 consecutive measurement days. Parents performed daily tympanometry at home. The study physician performed tympanometry and otoscopy at study entry, after 3 and 7 days, and then weekly until both ears were healthy. The main secondary outcome measures were the time to normal otoscopy findings and the proportion of children without persistent MEE at 14 days and 2 months.
Middle ear effusion disappeared 2.0 weeks (13.7 days) earlier (P = .02) in the antimicrobial group (mean time, 2.7 weeks; 95% CI, 1.7-3.7) than in the placebo group (4.7 weeks; 95% CI, 3.6-5.7). Normal otoscopy findings were observed 1.4 weeks sooner in the antimicrobial group than in the placebo group (P = .02). On day 14, 69% of children in the antimicrobial group and 38% in the placebo group had normal tympanometry findings (number needed to treat, 3.2; 95% CI, 2.0-10.5). On day 60, 2 children (5%) in the antimicrobial group and 10 children (24%) in the placebo group had persistent MEE (P = .01).
Conclusions and Relevance
Antimicrobial treatment effectively reduced the duration of MEE and possible concomitant hearing impairment in children with AOM. Antimicrobial treatment also reduced the risk for persistent MEE.
clinicaltrials.gov Identifier: NCT01244581
The antimicrobial treatment of acute otitis media (AOM) has proved effective in randomized clinical trials (RCTs); however, its effect on relieving symptoms has been modest.1- 4 It has been suggested that the antimicrobial treatment of AOM should be limited to children with significant symptoms.1 In addition to relieving symptoms, the antimicrobial treatment of AOM could be important in alleviating impaired hearing. Conductive hearing loss due to middle ear effusion (MEE) in early childhood could result in poorer speech development and cognitive skills later at school age, although in RCTs, prompt tympanostomy tube placement did not improve the later cognitive outcome of children with chronic MEE.5- 9
The effect of antimicrobial treatment of AOM on the duration of MEE and concomitant hearing impairment is not known.2 In most RCTs, the follow-up using tympanometry or otoscopy has been performed at weekly or monthly intervals after the initial episode.3,10- 15 Thus, the lack of evidence showing that MEE disappears faster after antimicrobial treatment could result from the inaccurate measure of MEE or new respiratory tract infections and new development of effusion in the middle ear after the initial episode. To improve the measurement accuracy, the middle ear status of children attending a trial should ideally be examined daily.
Because tympanometry is a reliable tool for determining the amount of MEE16,17 and hearing impairment,18 we designed a placebo-controlled RCT of the antimicrobial treatment of AOM and trained parents to perform daily tympanometry with a handheld device at home. With this approach, we aimed to clarify whether the antimicrobial treatment of AOM reduces the duration of MEE and concomitant hearing impairment.
This was a randomized (1:1), double-blind, placebo-controlled, parallel-group trial in children between 6 months and 15 years of age conducted in Oulu, Finland (Figure 1). The ethics committee of the Northern Ostrobothnia Hospital District, Finland, found the study protocol ethically acceptable. Only children whose parents provided written informed consent participated in the trial. Patients were recruited from September 14, 1999, to January 4, 2000 (31 children); October 10, 2005, to December 16, 2005 (12 children); and September 22, 2009, to June 4, 2012 (41 children), from among children in day care centers attending an AOM prevention trial at the Department of Pediatrics, Oulu University Hospital, Finland,19 and children visiting the City of Oulu Health Care Center and Mehiläinen Pediatric Private Practice, Oulu, Finland.
Eligible participants were 6-month-old to 15-year-old children with an AOM episode. Only children with acute symptoms of respiratory infection and/or ear-related symptoms and signs of tympanic membrane inflammation together with MEE detected in pneumatic otoscopy performed by a study physician were eligible for the trial.20 The otoscopists were validated against the tympanometry and tympanocentesis. Exclusion criteria were ventilator tubes, AOM complication, amoxicillin allergy, Down syndrome, congenital craniofacial abnormality, and immunodeficiency. The families of 120 children with AOM were invited to participate when a portable tympanometer (MiniTymp 1 or 2, Welch Allyn) was available (Figure 1). First, 31 children underwent tympanometry screening prior to the study; children who had normal tympanometry findings at the screening and later developed symptomatic AOM during a respiratory tract infection were eligible.19 In addition, 53 children were enrolled from the outpatient clinics and accepted into the study provided they had no history of AOM diagnosis within the previous month and had developed symptomatic AOM during a respiratory tract infection (Figure 1).
The children were randomly allocated to receive either an oral mixture of amoxicillin-clavulanate for 7 days (40 mg/kg of amoxicillin per day divided into 2 daily doses) or a placebo mixture. A computerized randomization list was created in blocks of 4 and was kept in the pharmacy (Oulu University Hospital Pharmacy and Oulun Vanha Apteekki Pharmacy), which delivered the study drugs to the families according to the consecutive study number. Both product bottles were indistinguishable and the dosing was similar in both groups. The placebo mixture was flavored and sweetened to resemble the taste of the amoxicillin-clavulanate mixture. The children, their families, and all participants of the study group were blinded to the treatment group until the data entry and checking were completed for all children.
The primary outcome measure was the time to the disappearance of MEE as defined by a normal tympanogram finding (A curve) from both ears on 2 consecutive measurement days either at home or at the study clinic. Analysis was performed per child. If tympanometry was not successful, the next available tympanogram was used in the analysis.
Secondary outcome measures were the time to improved tympanogram findings (ie, A or C curve) from both ears and time to normal pneumatic otoscopy or otomicroscopy findings from both ears. The proportions of children with persistent MEE on days 7, 14, and 60 were compared between the groups. The disappearance of pain and fever, the use of pain medication, and data on possible adverse effects of antimicrobial treatment were recorded and compared between the groups.
The study physician examined the children with pneumatic otoscopy or otomicroscopy and tympanometry at study entry, after 3 and 7 days, and then weekly until both ears were healthy according to pneumatic otoscopy or otomicroscopy. In addition to the scheduled visits, families were encouraged to contact the study physician during office hours or the emergency department of the Department of Pediatrics, Oulu University Hospital, at any time if the study participant experienced severe symptoms. In this case, either the study physician or an emergency department physician stopped administering the study drug and began the rescue treatment with amoxicillin-clavulanate in a nonblinded manner (Figure 1). If the child still had MEE after 60 days, the follow-up at the study clinic was stopped and the children with persistent MEE at that time were referred to the Department of Otolaryngology, Oulu University Hospital, for clinical examination after 1 month. Parents were advised to use a symptom sheet diary throughout the follow-up to collect daily data on other symptoms (fever >38.0°C, ear ache, cough, rhinitis, throat ache, vomiting, diarrhea, conjunctivitis, sleeping difficulties, and eating difficulties), as well as the number of study drug doses and other medications administered.
Each family received a handheld tympanometer for daily follow-up at home. Families were trained to perform tympanometry and instructed to repeat the measurement if they obtained a flat curve, to save and print the results of the tympanometry with a printer connected to the device, and to mark the measurement dates on the printed sheets of paper. Tympanometry was performed daily at home for up to 14 days or until a study physician declared both ears healthy according to pneumatic otoscopy or otomicroscopy findings.
Tympanometry is the measurement of acoustic admittance as a function of ear canal pressure and the resulting graph is a tympanogram. The main variables for detecting middle ear disease are static admittance (SA; ie, the ease with which acoustic energy flows into the middle ear) and tympanometric peak pressure (TPP; ie, the position of the tympanometric peak [the height of SA] on the pressure axis). Based on our validation and our previous studies,16,17,21 the tympanograms were classified as an A curve (SA ≥ 0.2 mmho; TPP, −200 to +100 daPa), a B curve (SA < 0.2 mmho), a C curve (SA ≥ 0.2 mmho; TPP < −200 daPa), a positive pressure curve (SA ≥ 0.2 mmho; TPP > +100 daPa), and an undefined curve (noisy measurement or other technical problem). Tympanogram with an A curve (normal finding) is highly specific (ie, the presence of normal tympanometry findings almost eliminates the possibility of MEE). Tympanogram with a B curve (abnormal finding) associates with MEE but can be due to other reasons such as stiffness of tympanic membrane. Tympanogram with a C curve (slightly abnormal finding) can result from the presence of MEE or low middle ear pressure. Two study physicians independently ranked the printed tympanograms. Interrater agreement was very good (κ = 0.95). If the 2 study physicians disagreed, the jury of the entire study group determined the classification. At the study clinic, the study physician recorded the possible reasons for unsuccessful tympanometry. The success rate of home tympanometry measurements was based on the proportion of successful tympanograms received of all intended measurements.
Sample size was calculated using sample-size calculation for survival analysis. The calculations were based on the median duration of MEE (7.5 days) evaluated with daily tympanometry in our previous study.21 We regarded longer median duration of MEE in the placebo group as clinically significant if the difference in median survival times was 7 days (7 days in the amoxicillin-clavulanate group and 14 days in the placebo group). Each group required 38 children for a type I error of 0.05 and a type II error of 0.20 (power, 80%). To achieve this in the final comparison, we decided to recruit 84 participants.
We used Kaplan-Meier survival analysis to analyze the durations of MEE and ear pain and the log-rank test to test the differences between the groups. We used a standardized normal deviate test to compare the proportions of the children with resolved MEE between the amoxicillin-clavulanate group and the placebo group, and we calculated the number needed to treat based on the absolute differences of the proportions between the study groups as well as 95% CIs for the differences. All analyses were performed in the intention-to-treat population (Figure 1). All children provided follow-up time until their ears were healthy or their parents terminated the study visits. Analyses were carried out with IBM SPSS Statistics version 20 and StatsDirect software 2.7.8.
A total of 84 children underwent randomization (Figure 1). The clinical characteristics of the participants at study entry were similar across the study groups (Table 1). One child in the amoxicillin-clavulanate group and 3 children in the placebo group prematurely discontinued the follow-up (Figure 1). In total, 15 children in the amoxicillin-clavulanate group and 11 children in the placebo group developed new symptoms of viral respiratory tract infection (after being asymptomatic at least for 2 days) before the MEE was resolved. The proportions of the study drug doses administered as planned were 96% in the amoxicillin-clavulanate group and 95% in the placebo group. The mean (SD) numbers of tympanograms were 18.3 (6.4) per child for each ear in the amoxicillin-clavulanate group and 20.7 (7.3) in the placebo group. Tympanometry proved successful in 95.8% of all examinations at the study clinic and in 86.5% of examinations at home. The child’s anxiety or crying was the main cause of failure of the examination at the study clinic.
The duration of MEE was shorter in the amoxicillin-clavulanate group than in the placebo group. Middle ear effusion disappeared 2.0 weeks earlier (P = .02) in the antimicrobial group (mean time to primary outcome, 18.9 days; 95% CI, 12.1-25.8) than in the placebo group (mean time to primary outcome, 32.6 days; 95% CI, 25.0-40.2) (Table 2, Figure 2A). The mean duration of MEE per AOM episode decreased by 8 days among children younger than 2 years of age, by 20 days among children 2 to 6 years of age, and by 1 day among older children (Table 2). The median time to MEE disappearance was 8.0 days (95% CI, 5.3-10.7) in the amoxicillin-clavulanate group and 29.0 days (95% CI, 3.3-54.7) in the placebo group.
Normal otoscopy findings appeared 9.7 days sooner (P = .02) in children treated with amoxicillin-clavulanate than in those in the placebo group (Table 2, Figure 2C). When combined tympanometry and otoscopy findings were used in the analysis, MEE resolved 11.5 days (P = .002) sooner in the antimicrobial group than in the placebo group (Table 2, Figure 2B).
On days 7 and 14 after administration of the study drug, the proportion of children with resolved MEE was greater among children who received amoxicillin-clavulanate than among those who received placebo (Table 3). The absolute difference in the percentage of children with resolved MEE between the study groups was 31% on day 14; 3.2 children (95% CI, 2.0-10.5) required treatment with amoxicillin-clavulanate to prevent 1 child from exhibiting abnormal tympanometry findings (Table 3). Two children (5%) treated with amoxicillin-clavulanate had persistent MEE (≥60 days), and 10 children (23%) in the placebo group had persistent MEE at the end of the trial (Figure 1) (95% CI of the difference, 4.5-34.7; P = .01). To prevent persistent MEE at 2 months in 1 child, 5.3 children required treatment with amoxicillin-clavulanate (95% CI for number needed to treat, 2.9-22.2).
The mean (SD) times to disappearance of the ear ache were 2.2 (2.2) days in the amoxicillin-clavulanate group and 3.2 (2.3) days in the placebo group (95% CI of the difference, −2.0 to 0.13 days; P = .08). The proportion of the children with ear ache differed statistically 5 days after administration of the study drug; at that time, none of the children in the amoxicillin-clavulanate group experienced ear pain, whereas 17% in the placebo group did (95% CI of the difference, 7.4-33; P = .004). The number of children requiring antimicrobial treatment to prevent 1 child from experiencing ear pain on day 5 was 5.8 (95% CI, 3.05-13.5). Differences in the dosage or duration of the pain medication between the groups were not statistically significant nor were differences in the disappearance of fever statistically significant.
Amoxicillin-clavulanate was administered as a rescue treatment for 1 child in the placebo group owing to severe symptoms (Figure 1). None of the children developed mastoiditis or tympanic perforation. Five children (12%) in the amoxicillin-clavulanate group and none in the placebo group had diarrhea during the first 10 days. All reported cases of diarrhea resolved within 3 days without treatment. Four children (9.5%) in the amoxicillin-clavulanate group and 2 children (4.8%) in the placebo group experienced vomiting within 10 days of study entry. None of the children had a rash or oral thrush during treatment.
Middle ear effusion disappeared significantly earlier in children who received amoxicillin-clavulanate than in children who received placebo in our study. In children receiving antimicrobial treatment, the mean duration of MEE was reduced by 2 weeks and the median duration of MEE by 3 weeks per AOM episode. To prevent 1 child from exhibiting abnormal tympanometry findings at 2 weeks, 3 children required antimicrobial treatment. Thus, antimicrobial treatment effectively reduced the duration of MEE and possible concomitant hearing impairment in children with AOM.
Our findings contradict those of several RCTs and those of 2 meta-analyses, which concluded that antimicrobial treatment has no impact on the duration of MEE evaluated either by tympanometry or otoscopy.1,2,10- 14 The most likely explanation for this discrepancy is that we obtained daily information on middle ear status. In this way, we had an accurate measure of the duration of MEE and avoided including new respiratory tract infections and the subsequent development of new MEE in the study population. In addition, the proportion of children with chronic MEE at study entry was low in our study. Thus, we were able to determine the benefit of antimicrobial treatment in the disappearance of MEE during 1 AOM episode.
Our results are important because MEE impairs hearing, which may cause long-term harm to later linguistic or other cognitive skills. Among children with MEE, flat tympanograms reportedly correspond to hearing levels ranging from 20 to 50 dB,22 whereas a peaked tympanogram eliminates the risk for hearing impairment.23 In a meta-analysis of prospective studies, both receptive and expressive language were more impaired in children who had previously had otitis media with effusion than in control participants.24 However, in randomized trials, prompt tympanostomy tube placement did not markedly improve the later cognitive outcome of children with chronic MEE over that of children who underwent the procedure later.7- 9
The recommended treatment of otitis media with effusion after an acute infection episode is a watchful waiting period of 3 months, with interval visits to a physician to document persistent effusion.25 In unresolved cases, hearing should then be examined and surgery should be considered.25 Approximately 700 000 myringotomies with the insertion of a tube and more than 100 000 adenoidectomies are performed annually in the United States in children younger than 15 years of age.26 Our study was not designed to estimate the difference of follow-up visits and subsequent surgery owing to chronic otitis media with effusion between the study groups. However, we are concerned that limiting the antimicrobial treatment of AOM may increase the need for subsequent follow-up visits and possible surgery because this study showed that the risk for persistent MEE at 2 months among children in the placebo group was 5-fold higher than in the antimicrobial group.
In young children and infants with AOM, antimicrobial treatment appears to have a modest effect on the proportion of abnormal otoscopy findings at 1 week4 and on the rate of remaining MEE at 3 weeks.3 However, in our study, the magnitude of the effect was greater, which may be owing to the older age of our study population. Pneumococcal resistance to penicillin is low in Finland.27 Consequently, the antimicrobial dosing appropriate for our setting may be insufficient for settings in other countries. Most of the study participants had not received pneumococcal conjugate vaccine, which can influence the spectrum of otopathogens.28 The optimal duration of antimicrobial treatment was not studied in this trial. We evaluated the efficacy of 1 week of antimicrobial treatment. Although the accuracy of tympanometry in detecting middle ear fluid is good,16,29 few studies have used handheld tympanometry for daily surveillance at home.21 Our approach was feasible as most of the home measurements were successful.
We conclude that the antimicrobial treatment of AOM in children is beneficial because it significantly reduces the duration of MEE. Thus, hearing impairment due to AOM can likely be resolved faster in children treated with antimicrobials.
Corresponding Author: Terhi Tapiainen, MD, PhD, Department of Pediatrics, Oulu University Hospital and University of Oulu, PO Box 5000, FIN-90014, Oulu, Finland (email@example.com).
Accepted for Publication: November 20, 2013.
Published Online: May 5, 2014. doi:10.1001/jamapediatrics.2013.5311.
Author Contributions: Dr Tapiainen and Ms Pokka had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Tapiainen, Kujala, Renko, Kontiokari, Kristo, Alho, Uhari.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Tapiainen, Kujala, Koivunen, Kontiokari, Kristo, Pokka, Uhari.
Critical revision of the manuscript for important intellectual content: Tapiainen, Kujala, Renko, Koivunen, Kristo, Pokka, Alho, Uhari.
Statistical analysis: Tapiainen, Pokka, Uhari.
Obtained funding: Tapiainen, Uhari.
Administrative, technical, or material support: Renko, Uhari.
Study supervision: Koivunen, Kontiokari, Alho, Uhari.
Conflict of Interest Disclosures: None reported.
Funding/Support: This work was financially supported by the Pediatric Research Foundation, Finland, and Medical Research Foundation, Finland.
Role of the Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: Leevi Luotonen, MD, Department of Otolaryngology, Oulu University Hospital, was the main study physician from 1999 to 2001 and died in 2001.