Chapter  55:  Diagnosis, Natural History, and Late Effects of Otitis Media With Effusion Volume 1: Evidence Report and Evidence Tables

Overview

The purpose of this evidence-based report is to review the evidence on the natural history of otitis media with effusion (OME), the impact of otitis media on long-term speech and language development and on hearing, and the operating characteristics of various methods of diagnosing otitis media with effusion. OME is defined as “fluid in the middle ear without signs or symptoms of ear infection.” The evidence compiled in this report is intended to aid clinicians, health care provider organizations, and others to develop clinical practice guidelines or medical review criteria for OME. The report will also identify areas for future research.

Reporting the Evidence

Based on degree of importance (including level of controversy) and feasibility of answering the question, the Technical Expert Panel limited the scope of this evidence report to four key questions. These questions address (1) the natural history of otitis media with effusion (OME), (2) the long-term effects of early-life otitis media, defined as positive otitis media history at less than three years of age, on speech and language development, (3) the long-term effects of early-life otitis media on hearing, and (4) the operating characteristics of various methods of diagnosing otitis media with effusion.

Methodology

A 12-member Technical Expert Panel that consisted of clinical experts, a consumer, and a representative of a managed care organization convened to provide the following assistance:

• advise the project in the ranking of proposed key questions and influencing factors

• guide the development of the scope and definition of OME

• advise in development of the search strategy, and

• review and comment on the plan of analysis.

The Technical Expert Panel and project staff developed a literature search strategy. Project staff searched MEDLINE (1966-January 2000), the Cochrane Library (through January 2000), and EMBASE (1980-January 2000). Additional articles were identified by review of reference lists in proceedings, published articles, reports, and guidelines.

The MEDLINE search strategy used both controlled vocabulary MeSH (Medical Subject Headings) terms and keywords to ensure that all relevant citations were retrieved. The strategy included search terms for otitis media with effusion combined with search terms for natural history, speech and language development, hearing, and diagnosis.

The otitis media module included otitis media, otitis media with effusion, suppurative otitis media, allergic otitis media, fluid ear, glue ear, middle-ear effusion, mucoid otitis media, nonsuppurative otitis media, secretory otitis media, and serous otitis media.

The natural history terms included natural course, natural history, placebo, placebos, resolution, self-limited, self limiting, and untreated, as well as a variety of terms for spontaneous resolution.

The speech and language module included speech and language, speech and language disorders, child language, communication, communication disorders, language development and tests, voice, and voice disorders.

The hearing module included hearing and hearing disorders, hearing aids and tests, and the text word hearing.

The diagnosis module used diagnosis and diagnostic techniques and procedures, as well as the text words audiometry, diagnosis, diagnostic, otoscopy, and tympanometry.

Two physicians or one physician and one health services researcher independently screened all titles and/or abstracts for potential inclusion, evaluated the quality of the articles, and abstracted data from full-length articles onto pre-designed forms. The selection criteria included human studies that addressed a key question about OME in children. Excluded were case reports, editorials, letters, reviews, practice guidelines, non-English language publications, and studies on patients with immunodeficiency disorders or craniofacial anomalies, including cleft palate.

For the natural history question, we used only prospective cohort(s) studies on untreated subjects from which outcome data were abstractable for children up through age 12 years. For the speech and language and hearing questions, we used only prospective cohort studies that fulfilled the following criteria: the degree of OME was determined during the first three years of life, upper age limit was 22 years, the degree of OM was graded in some way, and the outcome was measured when the child was older than age three years. For the diagnostic methods question, we used only prospective studies on children up through 12 years of age that fulfilled four criteria: the diagnostic procedure of interest was performed within 24 hours of the reference standard, was not an algorithm or combination of multiple diagnostic procedures, used one of the acceptable reference standards specified in the scope, and produced abstractable data.

The first step of all analyses was to obtain a distribution of studies stratified by the population characteristics, type of outcome measures, and non-treatment factors. This step provided us with an overview of the emphasis of past research in this area and an opportunity to identify gaps and areas for future research.

In strata with more than three studies, we performed a meta-analysis for a pooled random effects estimate of an outcome with 95% confidence intervals. In addition to deriving the pooled estimate, we evaluated the heterogeneity of the study outcomes. For the evaluation of diagnostic methods, we estimated the sensitivity, specificity, and positive and negative predictive values for each diagnostic procedure compared to a particular reference standard with three or more studies.

This evidence report was reviewed by the Technical Expert Panel as well as an 18-member peer review panel that consisted of content experts, consumers, representatives of managed care organizations, an expert in pediatric pharmacology, and methodologists. All comments received from these individuals were reviewed and acted upon appropriately.

Findings

Limitations of the Literature

• Natural History of OME: Literature on the natural history of otitis media with effusion was difficult to interpret because of its generally poor quality, the lack of control for therapeutic interventions, the inability to distinguish persistent from recurrent OME due to the length of follow-up intervals, and the varied criteria for continued follow-up from exam to exam. Differing definitions of OME resolution and diagnostic methods made comparison difficult. Few studies considered the child or the episode as the unit of analysis, included younger children, or assessed types of OME other than newly diagnosed OME of unknown duration. In addition, few studies addressed the possible effects of influencing factors on OME resolution.

• Early-Life OM and Long-Term Speech and Language Development: The literature on the long-term effects of early-life otitis media on speech and language development diverged considerably with respect to methodology. As a result, findings could not be combined easily.

• Early-Life OM and Long-Term Hearing: Although the literature on the long-term effects of early-life otitis media on hearing was abundant, few studies used a prospective cohort study design. Because of the limited nature of this evidence and because the rate of intervention is highly dependent on the threshold hearing level adopted, the findings of this analysis should be applied with caution.

• Diagnostic Methods for OME: Nine comparisons of diagnostic methods enabled derivations of pooled estimates of diagnostic accuracy. However, more comparisons could not be made, including those that would have evaluated clinical signs and/or symptoms, air and/or bone threshold audiometry, binaural micro-tympanoscopy, and non-pneumatic otoscopy. Diagnostic methods that use algorithms or aggregated scorings are important but are not included in this evidence assessment.

Future Research

Future research on the natural history of otitis media with effusion must focus on improvement of study quality and establishing the effect of OME on long-term outcomes such as speech, language, and hearing. In particular, control of therapeutic interventions during the study and the distinction between OME persistence and recurrence need to be addressed. Adopting a less restrictive definition of non-intervention might simplify the analysis of studies of the natural history of OME. In addition, researchers, in conjunction with clinicians, should agree upon standard procedures for follow-up, including intervals of follow-up, definition of OME resolution, and diagnostic methods, so that resolution rates are indeed comparable. Future research must consider the child as the unit of analysis, since the outcomes of ultimate interest, such as speech, language, and hearing, are functional requirements of a child, not an ear. More research is needed on the role of influencing factors on the natural history of OME, so that the clinician on a particular day in a particular setting can make a better decision when assessing a particular child with particular characteristics.

Evaluation of long-term effects of early-life otitis media on speech, language, or hearing requires a coordinated systematic approach that uses a rational conceptual framework. Such an approach should address the risk factors, interventions, and outcome measures in an integrated fashion. The definition, classification, and type and unit of measure should be developed by a team of experts with the goal of standardizing definitions and approaches. Literature on findings should report both univariate and multivariate findings to enhance understanding of the patient and study characteristics and to allow pooling of data. An integrated approach is also important for the evaluation of diagnostic methods. Such an approach will provide guidance for future studies. Future studies of diagnostic assessments of OME also should consider cost-effectiveness analysis, which can take into account the variable proficiency of clinicians in performing pneumatic otoscopy as well as the consequences of testing and patient preferences. Cost-effectiveness analysis will lead to a more informed decision on the best diagnostic method for OME.

Purpose

The purpose of this evidence-based report is to review the evidence on the diagnosis, natural history, and late effects of otitis media with effusion (OME) on long-term speech, language, and hearing. The evidence compiled in this report is intended to aid clinicians, health care provider organizations, and others to develop clinical practice guidelines or medical review criteria for OME. The report will also identify areas for future research. Despite the relatively recent publication of the 1994 Otitis Media with Effusion in Children guideline (Stool, Berg, Berman, et al., 1994), the technical experts believed that OME remained a topic worthy of evidence-based inquiry due to the continued controversy over the care of children with OME and the potential availability of new information.

Scope of Work

The technical experts initially proposed 20 questions that addressed 10 broad areas related to the diagnosis and treatment of otitis media with effusion: (1) allergens, (2) natural history, (3) speech and language, (4) diagnostic methods, (5) surgical interventions, (6) hearing, (7) antibiotics, (8) steroids, (9) antihistamines and decongestants, and (10) alternative or complementary therapies. These twenty questions were ranked based on the following criteria: (1) degree of potential impact on OME outcomes and on future guideline development and (2) the feasibility of answering the question within the one year time frame and the availability of new information in the literature. The scope of this report covers the four highest ranked questions: (1) the natural history of otitis media with effusion (OME), (2) the long-term impact of early-life otitis media on speech and language, (3) the long-term impact of early-life otitis media on hearing, and (4) the accuracy of methods of diagnosis of otitis media with effusion.

Definition

The definition of otitis media has been a complicated issue. Ben H. Senturia, quoted in Bluestone (1999), stated that “In the past, there has been a confusion of terms, in part because of a failure to distinguish conceptually between the disease process, otitis media, and one of the manifestations of that disease process, namely otitis media with effusion. Otitis media is dynamic and at any one time should be considered a single point in a continuum of the disease process.”

Recent comments on the definition of OME point to some of the complex issues involved:

1. The OME guideline (Stool, Berg, Berman et al., 1994) defined OME as “fluid in the middle ear without signs or symptoms of ear infection.” The guideline listed the following synonyms for OME: serous otitis media, secretory otitis media, allergic otitis media, catarrhal otitis media, nonsuppurative otitis media, mucoid otitis media, secondary otitis media, hydrotubotympanum, exudative catarrh, tubotympanitis, tympanic hydrops, glue ear, fluid ear, middle ear effusion, and tubotympanic catarrh.

2. Bluestone (1999) defined otitis media with effusion (OME) as “an inflammation of the middle ear with liquid collected in the middle-ear space. The signs and symptoms of acute infection are absent and there is no perforation of the tympanic membrane.” He stated that middle-ear effusions can be acute (< 3 weeks), subacute (3 weeks to 3 months), or chronic (> 3 months). He also stated that researchers should precisely define OME.

3. Patterson and Paparella (1999) agreed that the different forms of otitis media (OM) are “interrelated and occur in a continuum.” They recognized OME as one of the three major forms of OM, the other two being chronic OM (active or inactive) and silent OM. They classified OME as serous OM, purulent OM, and chronic OM. They divided serous OM into acute serous OM, chronic serous OM, and mucoid OM.

4. Jung and Hanson (1999) agreed that OM consists of various stages. Though they viewed purulent otitis media, serous otitis media, and mucoid otitis media as different stages, they considered OME to encompass all three, except the early stages of acute otitis media (AOM).

5. Paradise (1995) also agreed that “AOM and OME constitute elements in an otitis media disease spectrum, that there often is a transition zone between them and that the two conditions sometimes may be indistinguishable from each other diagnostically.”

For this evidence report, the Technical Expert Panel decided to use the definition used in the OME Guideline (Stool, Berg, Berman et al., 1994): “fluid in the middle ear without signs or symptoms of ear infection.”

Diagnosis

Various methods have been proposed for the diagnosis of OME. The OME guideline panel drew several conclusions regarding diagnosis of OME (Stool, Berg, Berman et al., 1994). They recommended the use of pneumatic otoscopy as the primary diagnostic method with tympanometry as a confirmatory diagnostic method. These recommendations were based on limited scientific evidence and strong panel consensus and on limited scientific evidence and expert opinion, respectively. The OME guideline panel found no evidence linking the outcome of algorithms that combine the results of pneumatic otoscopy and tympanometry to the presence of middle-ear effusion. In addition, the panel believed that the evidence was insufficient to make any recommendation regarding the use of acoustic reflectometry in the diagnosis of OME. Finally, the panel decided not to make a recommendation on the use of tuning fork tests in the diagnosis of OME due to the lack of adequate studies. The OME guideline panel did not present any meta-analyses on diagnostic methods.

Pneumatic otoscopy is performed with a handheld unit that consists of a light source, a magnifying lens, and a speculum. The otoscope allows visual inspection of the tympanic membrane as well as the external ear canal. With the speculum securely in place, the degree of movement of the tympanic membrane in response to pneumatic pressure may be observed. Decreased tympanic membrane mobility in response to pneumatic pressure is believed to be related to the presence of middle-ear effusion as found in OME. (Carlson and Stool, 1999)

Tympanometry is performed by inserting into the ear a probe that emits a tone and measures the amount of sound energy reflected from the tympanic membrane as a function of ear canal air pressure. The instrument may or may not be handheld. The output of tympanometric measurement may be qualitative, that is, tympanogram patterns, or quantitative, for example static admittance, equivalent ear volume, tympanometric width, tympanometric peak pressure, or acoustic reflex. The flat or type B tympanogram is believed to be associated with the presence of middle-ear effusion. The type A tympanogram is believed to indicate normal middle-ear status. The relationship of the type C tympanogram to middle-ear status is less clear. (Carlson and Stool, 1999; Nozza, 1996)

The acoustic middle-ear muscle reflex, either ipsilateral or contralateral, may also be measured by acoustic emittance instruments and represents the contraction of the stapedius and tensor tympani in response to sound stimulation. Its absence may be related to the presence of middle-ear effusion depending on the clinical situation (Nozza, 1996).

Acoustic reflectometry is performed using a handheld instrument that measures the response of the tympanic membrane to a frequency-sweep sound spectrum. The spectral gradient angle, which is a function of the frequency and amplitude, may be related to middle-ear effusion presence. (Carlson and Stool, 1999; Nozza, 1996)

Evoked otoacoustic emissions are a measure of ear canal sounds that are generated in the cochlea. These sounds have the potential for clinical application (Nozza, 1996).

Audiometry measures hearing acuity, using behavioral or non-behavioral methods, at various sound frequencies. It is known that children may have decreased hearing in the presence of middle-ear effusion (Carlson and Stool, 1999). Although no “universal agreement” appears to exist regarding the definition of hearing loss, a hearing threshold no worse than 15 decibels (dB) is considered normal in children, and 20 dB may be considered normal in older children (Madell, 1999)

Epidemiology: Prevalence and Incidence

Accurate estimates of the prevalence or incidence of OME were not found, since published population-based estimates are not available on the specific diagnosis of OME. Data on office visits reported from the National Ambulatory Medical Care Survey (NAMCS) provide the best indication of prevalence and incidence of the disease, although nonsuppurative, suppurative, and unspecified otitis media were all grouped into the term otitis media, and OME was not separated from AOM in the analyses (Schappert, 1992;Schappert, 1996; Woodwell, 1997a; Woodwell, 1997b; Woodwell and Schappert, 1995). Gates (1996), commenting on the NAMCS data, stated, “for children it is probably safe to presume that AOME [AOM with effusion, i.e. AOM] is the principal disorder noted in these surveys.” The OME Guideline panel of the Agency for Health Care Policy and Research [presently the Agency for Healthcare Research and Quality (AHRQ)] estimated that 25 percent to 35 percent of the NAMCS visits for otitis media were for OME (Stool, Berg, Berman et al., 1994a).

Schappert (1992) reported on the 1975 to 1990 NAMCS data. Visits by patients younger than 15 years of age constituted 70.6 percent of all office visits with the principal diagnosis of otitis media in 1975, 78.9 percent of all office visits with the principal diagnosis of otitis media in 1980, 81.9 percent of all office visits with the principal diagnosis of otitis media in 1985, and 80.5 percent of all office visits with the principal diagnosis of otitis media in 1990. From 1975 to 1990, the percent of office visits with otitis media as the principal diagnosis increased among those less than 15 years of age; from 7.3 percent to 17.4 percent for children under 2 years old, from 10.4 percent to 18.1 percent for the 2-5 year olds, from 6.9 percent to 10.5 percent for the 6-10 year olds, and from 2.6 percent to 5.2 percent for the 11-14 year olds. The number of visits with a principal diagnosis of otitis media per 100 persons per year for the same time period (1975 to 1990) increased from 31.5 to 102.1 for children less than 2 years of age, 20.8 to 47.8 for those 2-5 years of age, 10.2 to 18.2 for those 6-10 years of age, and 3.3 to 8.0 for those 11-14 years of age.

Rosenfeld (1994) noted that about a quarter of OME cases are discovered incidentally during well-child examinations. About 60 percent of children would have OME by 2 years old and 80 percent before school entry. The Agency for Healthcare Research and Quality 1994 OME guideline reported that in one study of children 2 to 6 years old in group child care, 53 percent had at least one episode of OME during the first year of the study, 61 percent had at least one episode during the second year of the study, and 30 percent had recurrent OME (Stool, Berg, Berman et al., 1994).

NAMCS also stratified data by specific physician type. From 1975 to 1990, the percent of office visits with a principal diagnosis of otitis media increased for pediatricians from 8.1 percent to 14.3 percent, for general practitioners and family physicians from 1.3 percent to 3.5 percent, and for otolaryngologists from 12.8 percent to 20.2 percent.

Data for 1975 to 1990 were also stratified by age. In 1990, the number of visits with a principal diagnosis of otitis media per 100 persons per year among children younger than 2 years was 62.9 for pediatricians, 24.0 for general practitioners and family physicians, and 9.1 for otolaryngologists. In 1990, the number of visits with a principal diagnosis of otitis media per 100 persons per year among children 2 to 5 years old was 29.0 for pediatricians, 11.4 for general practitioners and family physicians, and 6.6 for otolaryngologists (Schappert, 1992).

The reports on the NAMCS data for 1993 to 1996 did not stratify by age (Schappert, 1996; Woodwell, 1997a; Woodwell, 1997b; Woodwell and Schappert, 1995). However, if the 1993 to 1996 data were similar to that in 1975 to 1990, it would be reasonable to conclude that the majority of these patients were younger than 15 years of age.

NAMCS office visit data for 1993 to 1996 generally support earlier data. Suppurative and unspecified otitis media was the third most frequently listed principal diagnosis in 1993, the sixth most frequent in 1994, the fourth most frequent in 1995, and the seventh most frequent in 1996 for ambulatory care visits to physician offices, hospital outpatient departments, and emergency departments (Schappert, 1996; Woodwell, 1997a; Woodwell, 1997b; Woodwell and Schappert, 1995). In 1996, visits for a principal diagnosis of otitis media and eustachian tube disorders occurred 82.8 percent of the time in physician offices, 5.3 percent in hospital outpatient departments, and 11.9 percent in emergency departments (Woodwell, 1997a).

The NAMCS also provided data on the duration of office visits for otitis media. The percent of visits for otitis media of duration 6-15 minutes increased between 1975 to 1990 from 64 percent to 78 percent and was associated with a decrease in visits less than six minutes from 24 percent to 13 percent (Schappert, 1992). In terms of surgical procedures, the rate of ambulatory surgery visits per 10,000 population for those younger than 15 years of age for otitis media and eustachian tube disorders was 86.9 in 1994 and 83.9 in 1995, based on 498,000 and 484,000 visits respectively ( Hall and Lawrence, 1997; Kozak, Hall, Pokras et al., 1997). In 1995, the number of myringotomy with tympanostomy tube placements reported by NAMCS was 521,000 for a rate of 90.2 such procedures per 10,000 children less than 15 years of age (Hall and Lawrence, 1997).

In general, the NAMCS data demonstrated the significance of otitis media—and by implication OME—based on the prevalence and incidence of the disease and the frequency and duration of visits and surgical interventions.

Burden of Illness Due to Otitis Media with Effusion

The treatment, complications and sequelae, and adverse effects of otitis media, including OME, are a substantial financial burden to the nation. Three estimates of the cost of otitis media (OM) are found in the literature (Gates, 1996b; Stool and Field, 1989; Stool, Berg, Berman et al., 1994). A fourth study reported the cost per episode of “persistent middle ear infection” (Berman, Roark, and Luckey, 1994). We have assessed the strengths and weaknesses of these four studies and presented our own estimate of otitis media costs in a previous evidence-based analysis (Takata, Chan, Shekelle, et al., in press).

Gates (1996b) placed the annual national total cost of acute otitis media at $3.15 billion for the 0- to 4-year old age group in an unspecified date in the early or mid-1990's and placed the cost of treatment of chronic otitis media with effusion at $1.854 billion dollars per year (Gates, 1996). Stool and Field (1989) put the national cost of otitis media at $2.4-3.4 billion in the 0- to 6-year old age group in an unspecified year presumably in the middle or late 1980's. Both Gates (1996b) and Stool and Field (1989) assumed prevalences of otitis media that were at variance with the only available national data on the utilization of care visits for otitis media. Stool, Berg, Berman, et al. (1994) presented the only estimate of the national cost of otitis media using a data-based estimate of costs per case. Using claims data from more than 100 health insurers, they estimated the overall average cost of treating a 2-year old child with OME in 1991 to be $1,330 and the national total cost of OME in 1991 to be $1.09 billion. Berman, Roark, and Luckey (1994) estimated the cost of treating persistent middle ear effusion in a 13-month old boy at $720-$1,372 under Colorado Medicaid reimbursement levels in 1992 and $1,265-$2,588 under private practice reimbursement rates. Not all these studies included the indirect cost of family caregiving services required due to a child having otitis media.

An attempt to provide an updated estimate on the cost of otitis media included insights into the cost of OME and chronic middle ear infection (Takata, Chan, Shekelle, et al., in press). This estimate was derived for the year 1995, referred to children under 18 years of age, and was based on reports of national, rather than regional, utilization for otitis media such as the National Ambulatory Medical Care Survey and the National Health Interview Survey. Using these national data sets, it was estimated that 2.22 million episodes of OME or chronic middle ear infection occurred in 1995. It was assumed that eighty percent of these episodes were unrelated to acute otitis media. Data from three sources (Stool, Berg, Berman et al., 1994; Berman, Rourke, and Lucky, 1994; and the U.S. Bureau of the Census, 1992,1996) were used to estimate direct, indirect, and total costs of $1,321, $490.25, and $1,811 for treatment of an episode of OME or chronic middle ear infection. For 1995, the total national cost of treating OME or chronic middle ear infection would stand at $4.02 billion.

Whether based on the four prior estimates of otitis media or OME cost or the more recent estimate, the economic burden of OME on the nation is large. Any effort to improve cost-effective care of OME will result in significant savings in national medical expenditures as well as improved quality of care provided to children with OME.

Nomination of Technical Experts

Eleven organizations were contacted for technical expert and peer reviewer nominations. They included the American Academy of Family Physicians (AAFP), the American Academy of Pediatrics (AAP), the American Academy of Otolaryngology-Head and Neck Surgery Foundation (AAO-HNS), the Ambulatory Pediatric Association (APA), the American Academy of Audiology (AAA), the American Speech-Language-Hearing Association (ASHA), the Society for Ear Nose and Throat Advances in Children (SENTAC), the National Association of Pediatric Nurse Associates and Practitioners (NAPNAP), the American Association of Health Plans (AAHP), the Foundation for Accountability (FACCT), and Family Voices.

Table 1. Technical Expert Panel

Table 1. Technical Expert Panel

Technical Expert	Area of Expertise	Affiliation/Location
Larry Culpepper, MD, MPH	Family Medicine	Boston Medical Center, MA
Douglas G. Long, MD	Family Medicine	Manchester Community Health Center, NH
Richard M. Rosenfeld, MD, MPH	Otolaryngology	SUNY Health Science Center Brooklyn, NY
Norman Wendell Todd, Jr., MD	Otolaryngology	Emory University, GA
Allan Lieberthal, MD	Pediatrics	Southern California Kaiser Permanente Medical Group, CA
Anthony Magit, MD	Pediatric Otolaryngology	Children's Hospital San Diego, CA
Jack Paradise, MD	Pediatrics	Children's Hospital, Pittsburgh, PA
Ross Miller, MD	Quality Management	CIGNA Health Care, CA
Joanne Roberts, PhD	Speech and Hearing	University of North Carolina, NC
Lisa L. Hunter, PhD	Audiology	University of Minnesota, MN
Linda Carlson, MS, RN, CPNP	Nurse Practitioner	Statesboro, GA
Fran Goldfarb, MA	Consumer	Family Voices, Los Angeles, CA

Topic Assessment and Refinement

A draft work plan for the topic assessment and refinement phase was mailed to the technical experts and representatives of our partners (the American Academy of Pediatrics, the American Academy of Family Practice, and the American Academy of Otolaryngology-Head and Neck Surgery) for review and comments together with a preliminary review that provided a summary of:

• Incidence and prevalence of otitis media with effusion, treatment and management alternatives, the characteristics and size of the affected populations, and the most affected practice settings and providers;

• The burden of illness associated with otitis media with effusion, including morbidity and mortality.

• Extent to which variation exists in practices associated with the prevention, diagnosis, treatment, or diagnosis and treatment of otitis media with effusion.

We reviewed and compiled four previously conducted evidence-based analyses on otitis media with effusion. Particular attention was given to the report of The Otitis Media Guideline Panel on “Managing Otitis Media with Effusion in Young Children” ( Stool, Berg,Berman et al., 1994) We distributed these reports to the panel of technical experts for review and for preparation of the initial telephone conference call during which assessment and refinement of the topics wereassessed and refined.

Identification of Key Questions

Table 2. Questions Suggested for Consideration in Evidence (more...)

Table 2. Questions Suggested for Consideration in Evidence Report

1.	What is the relative risk of developing OME in the child who has food or inhalant allergies compared to the child without food or inhalant allergies
2.	What is the natural history for various types of OME?
3.	What is the long-term level of speech and language development in infants and preschool children with untreated OME? What are the high risk groups?
4.	What is the accuracy of various diagnostic methods?
5.	When should conservative treatment (non-surgical) be considered a failure?
6.	What is the evidence on effectiveness of various diagnostic instruments in deciding on intervention for OME?
7.	What is the evidence regarding level of hearing decrease and whether unilateral or bilateral hearing decrease is an indication for intervention?
8.	What is the effectiveness of the use of hearing levels to decide on intervention for OME?
9.	Are antibiotics more effective than placebo in treating OME?
10.	Are steroids more effective than placebo in treating OME?
11.	Do antibiotics add an incremental benefit to steroids in treating OME?
12.	Are interventions for allergies (food or inhalant) more effective than placebo in treating OME?
13.	Are antihistamines and/or decongestants more effective than placebo in treating OME?
14.	Are tympanostomy tubes more effective than other interventions in treating OME?
15.	Is adenoidectomy more effective than other interventions in treating OME of greater than 3 months duration?
16.	Is tonsillectomy more effective than other interventions in treating OME of greater than 3 months duration?
17.	Is myringotomy more effective than other interventions in treating OME of greater than 3 months duration?
18.	Are alternative or complementary therapies more effective than other interventions in treating OME of greater than 3 months duration?
19.	Are prophylactic antibiotics more effective than other interventions in treating OME?
20.	What is the effectiveness of monitoring by pneumatic otoscopy, tympanometry, acoustic reflectometry with spectral gradient, and otoacoustic emissions to decide on intervention for OME?

1. importance, which included

   1. potential impact on OME outcomes and

   2. potential impact on development of future OME guidelines by the partner organizations; and

2. feasibility, which included

   1. possibility of conducting a literature search, review, and data synthesis in 6 months,

   2. availability of sufficient information (data) in the literature to answer the question, and

   3. if applicable, sufficient new information (data) available to affect the results of the last systematic review of the question significantly.

Table 3. Ranking of Potential Key Questions. Of the (more...)

Table 3. Ranking of Potential Key Questions. Of the 20 questions, each technical expert ranked the top 10 questions from 10 (highest priority) to 1 (lowest priority), (experts’ identification numbers were randomly assigned)

	Topic of question	Rank Total	E1	E2	E3	E4	E5	E6	E7	E8	E9	E10	E11	E12
1.	Food or inhalant allergies	8				5					3
2.	Natural history	78	10	9	10	9	7	9	4	2		6	7	5
3.	Speech and language development	97	8	10	4	8	9	5	8	5	10	10	10	10
4.	Accuracy of diagnostic methods	57			5	10	6	10	6	9		8	1	2
5.	Conservative treatment	39	9	5	6	4	1		3			1	9	1
6.	Diagnostic instruments vs intervention	38	7	8				6		9		8
7.	Level of hearing decrease	67.5		7		7	8	4	9	6.5	9		8	9
8.	Hearing levels and intervention	16.5		6	1			3		6.5
9.	Antibiotics versus placebo	51	6	4	9	6	5		1		7	5	2	6
10.	Steroids versus placebo	23					4		7	1	6		5
11.	Antibiotics and steroids	22	5	2	3		3				5		4
12.	Treatment for allergies vs placebo	2	1			1
13.	Antihistamines/decongestants vs placebo	10							10
14.	Tympanostomy tubes vs other interventions	51	3	1	8		10	2	2	4	4	3	6	8
15.	Adenoidectomy vs other interventions	32			7		2	1	5	3	2	2	3	7
16.	Tonsillectomy vs other interventions	2	2
17.	Myringotomy vs other interventions	5			2	3
18.	Alternative/complementary therapies vs other interventions	10				2		7			1
19.	Prophylactic antibiotics vs other interventions	15	4	3								4		4
20.	Effectiveness of diagnostic methods for monitoring	36						8		9	8	8		3

Note: Kendall Coefficient of Concordance = 0.36, p=0.0001.

Table 4. Comments from Technical Expert Panel on Potential (more...)

Table 4. Comments from Technical Expert Panel on Potential Key Questions

Potential Key Questions		Comments/Notes
1.	What is the relative risk of developing OME in the child who has food or inhalant allergies compared to the child without food or inhalant allergies?	Difficult to obtain evidence Doesn't seem to have adequate research base Assumed to be a minor player, but there could be a surprise here
2.	What is the natural history for various types of OME?
3.	What is the long-term level of speech and language development in infants and preschool children with untreated OME? What are the high risk groups?	Evidence incomplete Several new prospective studies are available This is probably the most important single question for justifying surgical intervention Available results inconsistent and/or contradictory. No definitive answers yet available
4.	What are the accuracy of various diagnostic methods?	Combine with 6 and 20 Probably not much data there Add binocular micro-tympanoscopy, MRI, and quantitative tympanometry to the list of methods of diagnosing OME. Binocular micro-tympanoscopy, which I do to every patient that I assess in the clinic, is using the surgical “operating” microscope to view each tympanic membrane. I think MRI should be the “gold standard” (see Swarts JD et al.: In vivo observation with magnetic resonance imaging of middle ear effusion in response to experimental underpressures. Ann Otol Rhinol Laryngol 104:522-528, 1995). To diagnose OME by looking at the tympanic membrane is traditional, but provides only a look at the “window on the middle ear”. To use fluid found at myringotomy as the “gold standard” assumes that either (1) no inhalational general anesthetic has been given to the patient, or (2) there is no effect of inhalational general anesthetic on the finding of fluid in that part of the mesotympanum when the myringotomy is done. Since practically, inhalational general anesthetic is usually administered to patients getting myringotomy in the US in 1999, I'm not comfortable with this first assumption. From clinical experience, I've very uncomfortable with the second assumption. MRI also affords an opportunity to measure the volume of the middle ear system (I'm including mastoid, mesotympanum, epitympanum and all the spaces normally containing “gas” in this definition of the middle ear system); smallness of volume of the middle ear system is a well-documented correlate of severity of otitis media condition, and also a correlate of the severity of the anatomic eustachian and skull base differences of otitis media.) [The MRI data would also help, I think, in answering ”Potential Key Question # 2.] Quantitative tympanometry is a promising technique. ANSI mandated that all new tympanometers (as of 1996, as I recall) provide quantitative information. Such quantitative information, [DeChicchis & Todd, unpublished data] to date show advantages over the qualitative A-B-C Jerger classification.
5.	When should conservative treatment (non-surgical) be considered a failure?	Unclear question Very broad category Unclearly written. What is the conservative treatment No definitive information on which to base an answer
6.	What is the evidence on effectiveness of various diagnostic instruments in deciding on intervention for OME?	Same as 4 Combine with #4 and #20
7.	What is the evidence regarding level of hearing decrease and whether unilateral or bilateral hearing decrease is an indication for intervention?	Minimal evidence Combine with 8 Pairs with question 3 No definitive information on which to base an answer
8.	What is the effectiveness of the use of hearing levels to decide on intervention for OME?	Same as 7 Combine with 7
9.	Are antibiotics more effective than placebo in treating OME?	Is this still an issue? Is this still an issue? Has it been addressed with meta-analysis already? Let's see if there is any new data out there
10.	Are steroids more effective than placebo in treating OME?	There wasn't enough data during the original panel. Is there now?
11.	Do antibiotics add an incremental benefit to steroids in treating OME?	This area showed some potential promise last time Should be “Do steroids add an incremental benefit to antibiotics”
12.	Are interventions for allergies (food or inhalant) more effective than placebo in treating OME?	Minimal evidence Inadequate research base
13.	Are antihistamines and/or decongestants more effective than placebo in treating OME?	If used need to indicate if there is a presence or absence of allergies Still an issue? Addressed in last OME guidelines
14.	Are tympanostomy tubes more effective than other interventions in treating OME?	Good to test the current recommendation to see if they still have validity Answers might vary depending on outcome measures
15.	Is adenoidectomy more effective than other interventions in treating OME of greater-than 3 months duration?	Need to indicate in what age group and if used combine with myringotomy The key is age of patient. Are there studies for younger children? Answers might vary depending on outcome measures Though I think adenoidectomy has some benefit to some patients in helping the resolution of their otitis media, none of the purported mechanisms make any sense to me. One purported mechanism is to decrease the “cesspool” of the nasopharynx; but, cystic fibrosis patients, who typically have the worse cesspools of any nasopharynges, have (on average) less otitis problems than the general population. Another purported mechanism is removing the mass of lymphoid tissue in the nasopharynx; but, I don't know of any data that size of adenoid tissue correlates with either the occurrence or severity of otitis media. Indeed, in my experience with a population who have one of the highest rates of otitis of any population, adenoid tissue is usually scant. I'd love to know the explanation by which adenoidectomy improves the course of otitis media in some children. (I suspect the explanation is scarring in the adenoid bed, that stabilizes the posterior lamina of the eustachian cartilage. If this is indeed the explanation, then it is one of the rare occasions that surgery benefits a patient by inducing scarring.)
16.	Is tonsillectomy more effective than other interventions in treating OME of greater-than 3 months duration?	Still an issue?
17.	Is myringotomy more effective than other interventions in treating OME of greater-than 3 months duration?	Still an issue
18.	Are alternative or complementary therapies more effective than other interventions in treating OME of greater-than 3 months duration?	Minimal evidence Inadequate research base? Probably won't find much, but the public will be clamoring for it
19.	Are prophylactic antibiotics more effective than other interventions in treating OME?	Still an issue
20.	What is the effectiveness of monitoring by pneumatic otoscopy, tympanometry, acoustic reflectometry with spectral gradient, and otoacoustic emissions to decide on intervention for OME?	Similar to 6 combined Combine with #6 and #4 Add binocular micro-tympanoscopy, MRI, and quantitative tympanometry to the list of methods of diagnosing OME. See my comments about ”Potential Key Question” 4.
	Epidemiology	The material on epidemiology thus far circulated omits mention of a key factor in predisposing infants and children to otitis media, namely, low socioeconomic status. For data and a discussion, see our report in Pediatrics 1997;99:318-333.
	Definition of AOM	I disagree with the definition of AOM as stated in your recent Definition section. As stated, the definition would call for a diagnosis of AOM in a child with middle-ear effusion and rapid onset of either irritability or fever. On the one hand, some infants with AOM have none of the 4 listed signs or symptoms, and the diagnosis is made on the basis of specific tympanic membrane findings--bulging and/or marked erythema--in addition to findings of middle-ear effusion. On the other hand, some infants with only OME rather than AOM present with rapid onset of fever and/or irritability that may be due the underlying viral respiratory tract infection. If such patients are assumed to have AOM, much unnecessary antibiotic will be prescribed. The point to be made is that diagnosis should not be linked to either the presence or the absence of signs or symptoms such as fever and irritability that are nonspecific. Ear pain, on the other hand, is reasonably specific. This issue was discussed in Commentary in Pediatrics 1995;96:712-715.

The four top ranking key questions were selected for consideration in the evidence report. After several revisions at the suggestions of the technical experts, the wordings of the four top ranking (key) questions are as follows:

Identification and Refinement of Causal Pathways, Study Populations, Practice Settings, and Target Audience

The project staff developed a causal pathway and a scope for each of the key questions. We distributed these documents to the panel of technical experts for review and for preparation for the initial telephone conference call. During this call, we assessed and refined the topics and discussed the proposed key questions, target condition, patient populations, clinical context, interventions, and outcomes of interest. The following characteristics of outcomes were proposed: 1) outcomes would be divided into short term and long term; 2) long term outcomes would consist of percent time with effusion, frequency of acute otitis media, hearing loss, speech and language performance, cognition, academic achievement, and other developmental outcomes; 3) duration of short term outcomes would be defined as four weeks or less or eight weeks or less; and 4) duration of long term outcomes would be defined as greater than one year.

During this first conference call, the Technical Expert Panel decided to use the OME Guideline definition of OME: “fluid in the middle ear without signs or symptoms of ear infection”. (Stool, Berg, Berman et al., 1994) Though the technical experts agreed on the definition of OME they could not agree on which signs or symptoms should be absent, i.e. what signs or symptoms differentiated OME from acute otitis media. During the first conference call, the technical experts advised us to avoid the use of the terms ‘acute’, ‘subacute’, or ‘chronic’ as descriptors of OME. Instead we should use the duration of OME, such as “under 3 months” versus “greater than or equal to 3 months” as descriptors.

Table 5. Overall Casual Pathway for OME

Table 5. Overall Casual Pathway for OME

Table 6. Casual Pathway for Key Question 1 on Natural (more...)

Table 6. Casual Pathway for Key Question 1 on Natural History

What is the natural history (spontaneous resolution rate over time without treatment) for:

•OME persisting after a discrete episode of acute otitis media,

•newly diagnosed OME of unknown duration (unilateral or bilateral)

•OME persisting for weeks or months (unilateral or bilateral)

•unilateral OME lasting 3 months or longer,

•bilateral OME lasting 3 months or longer?

Table 7. Casual Pathway for Key Question 2 on Speech (more...)

Table 7. Casual Pathway for Key Question 2 on Speech and Language

Table 8. Casual Pathway for Key Question 3 on Hearing (more...)

Table 8. Casual Pathway for Key Question 3 on Hearing

Table 9. Casual Pathway for Key Question 4 on Diagnostic (more...)

Table 9. Casual Pathway for Key Question 4 on Diagnostic Methods

Based on the causal pathways and the discussions during the two conference calls, we developed the scope for each of the four key questions. The scope specifically defined the disease entity, study population, practice settings (including provider type), time period in practice setting, exclusion factors, interventions, influencing factors, outcome measures, literature sources, language, and study design for each key question to be included in the evidence report. We conducted a second poll of the technical experts on each of these domains in which we sought their approval, disapproval, or recommendations for revision on each domain.Appendix B presents the version of the scope distributed to the technical experts for polling of their comments and approval. Appendix C presents the results of the polling of the experts' comments on the scope. In response to these comments, we further revised the key questions, causal pathways, and the various domains of the scope. We incorporated comments from technical experts to the extent possible, except those related to other domains, those that were obvious misunderstandings or misinterpretations, or those suggestions for deletions or additions that could be handled during the analysis phase of the project. Specifically, the project team retained several influencing factors recommended for deletion by some experts. The project team took note of the deletions and would later stratify the analysis, if possible, by the group of factors unanimously recommended by the experts versus those that were not. Such stratified analysis would depend on the number of studies that specifically address these factors. We revised the key questions according to experts' suggestions. We further revised the scope and reworded the key questions according to the final round of comments: the final version of the scope is included in Appendix D.

In preparation for supplemental analysis, we took polls to solicit the technical experts' opinion on the importance of the risk factors identified in the analytical framework. Specifically, we asked the experts the following questions for each factor:

1. Regarding Key Question 1: “Does this factor influence the natural history of OME?”

2. Regarding Key Question 2: “Does this factor have an independent effect on speech and language development separate from its effects on OME or unspecified OM?”

3. Regarding Key Question 3: “Does this factor have an independent effect on long-term hearing separate from its effects on OME or unspecified OM?”

4. Regarding Key Question 4: “Does this factor have an independent effect on the accuracy of a diagnostic method separate from its effects on OME or unspecified OM?”

The experts had a choice of responding “yes”, “no”, or “don't know.” For each such opinion, the experts were asked to indicate the basis of their opinion by choosing one or more of the following: “Judgment/Experience,” “Theoretical Construct,” or “Literature”.

Table 10. Ranking of Influencing Factors by Technical (more...)

Table 10. Ranking of Influencing Factors by Technical Experts, Key Question 1

Key Question 1: Natural History
Factors	Total “Yes” Based on Responses of 11 Technical Expertsa
total duration of OME (≥3 mos)	10
otitis prone (AOM)	9
number of previous OMEs	9
number of hours attending child care center	9
tobacco smoke exposure	9
season of the year	9
age at first OM	9
not breast-fed	9
Allergies	8
age of child	8
family history of OME	7
number of children in household	7
prior tubes	7
ethnicity/race	6
barotrauma challenges	6
prior adenoidectomy	5
socioeconomic status	5
laterality, unilateral versus bilateral	4
Gender	3
skill to diagnose (validated)	3
age of onset of previous OME	3
tympanometry	3
monitoring frequency	2
monitoring time	2
MRI	2
type of examiner	2
setting of care	1
parent/caregiver education	1
hearing level, conductive versus sensorineural	1
primary provider	1
acoustic reflectometry	1
Otoscopy	1
pneumatic otoscopy	1
parent/caregiver preference for treatment	1
developmental delay	0

a

11 technical expterts responded; 1 astained

Table 10a. Ranking of Influencing Factors by Technical (more...)

Table 10a. Ranking of Influencing Factors by Technical Experts, Key Question 2

Key Question 2: Speech and Language
Factors	Total “Yes” Based on Responses of 12 Technical Expertsa
developmental delay	11
quality of child care	10
hearing level, conductive versus sensorineural	10
parent/caregiver education	10
quality of parent-child interaction	10
socioeconomic status	8
laterality, unilateral versus bilateral	8
early intervention program	7
total duration of OME (≥3 mos)	7
Gender	6
number of children in household	6
duration of middle ear effusion	6
chronic illness of any type	6
number of hours attending child care center	5
number of previous OMEs	4
presence of active ear disease	4
ethnicity/race	3
tobacco smoke exposure	3
OM complications	3
child temperament	3
Allergies	2
ambient noise	2
age at first OM	1
not breast-fed	1
skill to diagnose (validated)	1
type of examiner	0
setting of care	0
recheck times	0
frequency of recheck	0
primary provider	0
tympanometry	0
acoustic reflectometry	0
pneumatic otoscopy	0
MRI	0
equipment	0
audiometry	0

Note:Items in bold were added in the second poll after the first poll.

Table 10b. Ranking of influencing Factors by Technical (more...)

Table 10b. Ranking of influencing Factors by Technical Experts, Key Question 3

Key Question 3: Hearing
Factors	Total “Yes” Based on Responses of 12 Technical Experts
OM complications	10
laterality, unilateral versus bilateral	7
hearing level, conductive versus sensorineural	7
developmental delay	6
presence of active ear disease	6
total duration of OME (≥3 mos)	4
number of previous OMEs	3
duration of middle ear effusion	3
chronic illness of any type	3
ambient noise	3
Allergies	2
age at first OM	1
ethnicity/race	1
socioeconomic status	1
quality of child care	1
early intervention program	1
tobacco smoke exposure	1
number of children in household	1
child temperament	1
Equipment	1
Audiometry	1
Gender	0
number of hours attending child care center	0
not breast-fed	0
parent/caregiver education	0
quality of parent-child interaction	0
skill to diagnose (validated)	0
type of examiner	0
setting of care	0
recheck times	0
frequency of recheck	0
primary provider	0
Tympanometry	0
acoustic reflectometry	0
pneumatic otoscopy	0
MRI	0

Note:Items in bold were added in the second poll after the first poll

Table 10c. Ranking of influencing Factors by Technical (more...)

Table 10c. Ranking of influencing Factors by Technical Experts, Key Question 4

Key Question 4: Diagnostic Tests
Factors	Total “Yes” Based on Responses of 11 Technical Expertsa
age of child	11
Otolaryngologist	6
nurse practitioner	5
Pediatrician	5
family physician	5
laterality, unilateral versus bilateral	5
Anesthetic	5
age at first OM	5
developmental delay	4
physician assistant	4
Others	3

a

11 technical experts responded; 1 abstained.

Literature Search

The Technical Expert Panel and project staff developed a literature search strategy. The literature search included the search of three databases: MEDLINE (1966-January2000), the Cochrane Library (through January 2000), and EMBASE (1980 -January 2000). We identified additional articles by review of reference lists in proceedings, published articles, reports, and guidelines.

The project librarian developed an overall search strategy for MEDLINE (Appendix G) that incorporated the input from the technical experts and followed the scope of the project. The MEDLINE database is produced by the U.S. National Library of Medicine and is widely recognized as the premier source for bibliographic coverage of biomedical literature. It encompasses information from Index Medicus, Index to Dental Literature, and International Nursing, as well as other sources of coverage in the areas of allied health, biological and physical sciences, humanities, and information science as they relate to medicine and health care, communication disorders, population biology, and reproductive biology. We searched the MEDLINE database for publications dating back to 1966. Further, we included articles in the English language only for the following reasons. First, our experience with our evidence assessment of the management of acute otitis media (Takata, Chan, Shekelle et al., in press; Chan, Takata, Shekelle et al., in press) demonstrated a low yield from non-English language publications (only two studies accepted out of 97 reviewed and both of theses were also published in English). Second, we needed to balance limited resources between reviewing non-English language literature and answering additional key questions. Since empiric evidence of the need to include non-English language literature in meta-analyses was mixed (Moher, Pham, Klassen et al., 2000) and reviewing non-English literature would be resource intensive, we chose to limit our scope to English language literature only.

The MEDLINE search strategy used both controlled vocabulary terms and keywords. The strategy was organized into modules or clusters of search statements. The main groupings included: otitis media with effusion (OME), mastoid, otitis media; natural history; speech and language; hearing; and diagnosis. These groupings corresponded to the key questions.

For the “otitis media with effusion” concept, both the controlled vocabulary term otitis media with effusion and text word were used. A variety of additional terms were used; such as allergic otitis media, fluid ear, glue ear, middle ear effusion, mucoid otitis media, nonsuppurative otitis media, secretory otitis media, and serous otitis media. For the “mastoid” concept, both the controlled vocabulary and the text word were used. The otitis media module included what is referred to as an “explode” of otitis media, which included the controlled vocabulary headings “otitis media”, “otitis media with effusion”, and “otitis media, suppurative.”

The “natural history” module combined “OME” or “mastoid” with a combination of text words and controlled vocabulary terms for natural history including “natural course”, “natural history”, “placebo”, “placebos”, “resolution”, “self limited”, “self limiting”, “untreated”, and a variety of terms for spontaneous resolution.

Both the “speech and language” module and the “hearing” module combined OME, or mastoid, or an explode of “otitis media” with the speech, language, and hearing concepts. The speech and language component used the controlled vocabulary terms for speech and language, speech and language disorders, child language, communication, communication disorders, language development and tests, voice, and voice disorders. In addition, the text words “speech” and “language” were added. The hearing module used the controlled vocabulary terms for hearing and hearing disorders and hearing aids and tests, as well as the text word “hearing”.

The “diagnosis” module combined OME or mastoid with a combination of text words and controlled vocabulary terms for diagnosis. In addition to the controlled vocabulary terms for diagnosis and diagnostic techniques and procedures, a number of text words were added for audiometry, diagnosis, diagnostic, otoscopy, and tympanometry.

We customized the search strategy initially developed for MEDLINE for EMBASE. EMBASE, the Excerpta Medica database, produced by Elsevier Science, is a major biomedical and pharmaceutical database that indexes over 3,800 international journals. EMBASE is one of the most widely used biomedical and pharmaceutical databases. The database currently contains over 6 million records, with more than 400,000 citations and abstracts added yearly. We searched the EMBASE database for citations dating back to 1980. For the search in the Cochrane Library, we used “otitis media” as the search term. The Cochrane Library contains several databases: (1) The Cochrane Database of Systematic Reviews, which contains Cochrane reviews published by the Cochrane Collaboration, an international organization dedicated to applying evidence-based-medicine principles to the review of important clinical topics; (2) The Cochrane Controlled Trials Register which is a bibliographic database of controlled trials; (3) The Database of Abstracts of Reviews of Effectiveness (DARE), which includes structured abstracts of systematic reviews that have been critically appraised by reviewers at the NHS Centre for Reviews and Dissemination in York and by other experts, for example from the American College of Physicians'; Journal Club and the journal Evidence-Based Medicine; and (4) The Cochrane Review Methodology Database which is a bibliography of articles on the science of research synthesis.

The Cochrane Library search yielded 666 titles/abstracts. The MEDLINE search resulted in 2,379 titles/abstracts. After eliminating duplicates, we retained 2,207 titles/abstracts for screening. The EMBASE search retrieved 1980 citations. After eliminating duplicates, we retained 327 for screening.

We conducted all searches in January 2000 and subjected a total of 3,200 titles/abstracts to screening by two physician reviewers. By merging and eliminating duplicates from the titles/abstracts from the three databases, we created a database of titles and abstracts.

EndNote software (EndNote Windows Version 3.0, 1st Edition. Niles Software Inc., Berkeley, CA) was used to keep a complete record of all titles/abstracts and identify duplications. This software stores, organizes, and tracks references by source (e.g. identified in MEDLINE), search strategy (date of search, index code specifying search criteria used), and a unique identification (UI) code for each article (assigned by the source used to find article). Electronic removal of duplicate citations was supplemented by manual cross-checking. In the event an article was identified by an expert panel member or through reference checking, the title and author of the reference were entered into MEDLINE through the Ovid search system (Ovid Technologies, Inc. 1998, Version: 7.8 Millennium source ID 1.3932.1.156.1.7, Revision: 1.303.2.8) to determine the UI. If a UI could not be found for the article, an alternate identification code was assigned.

EndNote assigned a record number to each new reference added to the master file. This number would not change once an article was added to the list and was used, in addition to the UI, to sort references for article retrieval and review.

Upon completing the literature search and duplicate checking, we exported the master list generated from EndNote to a Microsoft Excel spreadsheet for data export and analysis. We added codes including status of article retrieval, reviewer, and the results of the review.

Review of Retrieved Titles/Abstracts Against Screening Criteria

After retrieving of titles and abstracts from the literature search, two physician reviewers reviewed the abstracts against the inclusion/exclusion criteria to determine eligibility for inclusion in the evidence synthesis as defined in the scope and key questions. Titles/abstracts were not masked prior to review. A pre-designed screening form was used to record the reviews. A meeting was held to review the instructions for screening (Appendix H), including the use of the computerized data forms. The reviewers entered the screening results directly into the computer and forwarded the results electronically to our data analyst for processing. The screening results for each title/abstract were matched between the two reviewers, and discrepancies on inclusion or exclusion were resolved in conference calls among the two reviewers and the task order coordinator. The data analyst generated summary reports indicating those abstracts that passed the screening criteria and those that failed and the reasons for failure.

Table 11. Reason for Rejection of Titles/Abstracts (more...)

Table 11. Reason for Rejection of Titles/Abstracts at Initial Screening (N=3,200)

Reason Code	Reason for Rejection	Number (Percent) of Citations
R0	Written in non-English language	170 ( 5.3%)
R1	Case report/editorial/letter/clinical/overview/ practice guidelines/consensus statements	459 (14.3%)
R2	Non-human subjects	19 ( 0.6%)
R3	Study condition not OM	804 (25.1%)
R4	Age of study population >12 yearsa	57 ( 1.8%)
R5	Study population exclusively on any one of the following: Craniofacial defects, primary mucosal disorders, Immunodeficiencies, or Down or other genetic syndromes	15 ( 0.5%)
R7	Any key questions not addressed	697 (21.8%)
R8	Duplicate citation	9 ( 0.3%)

a

The age limit was later extended to 22 years of age for Questions 2 and 3.

We completed screening of the 3,200 titles/abstracts from the Cochrane Library, MEDLINE, and EMBASE. After resolution of 376 discrepant citations, 2230 (70%) were rejected and 970 were accepted for full article review. The reasons for rejection of the 2230 citations are presented in Table 11. We also screened the database provided to us by the American Academy of Pediatrics (AAP) from its recent review of the topic. Of a total of 1918 titles/abstracts screened against our database, we identified 477 duplicates from our ENDNOTE database, leaving 1441 records from the AAP files that required further screening by the two reviewers. The screening of the 1441 citations from AAP files identified 32 additional citations that required full article review.

The third source of reference material was the six proceedings of the International Symposium on Recent Advances in Otitis Media with Effusion from which we identified 159 additional citations for full article review.

The fourth source consisted of references in books and articles from which we identified 31 full articles for further review.

Retrieval and Review of Full Articles

The titles/abstracts identified for further review were forwarded to the library for full article retrieval. Libraries at both the Los Angeles County - University of Southern California Medical Center and the University of Southern California Health Sciences Campus were the primary sources of the articles. Those not found were retrieved through the Inter-Library Loan Program.

Because a large number of titles/abstracts had inadequate information for full evaluation, a secondary screening of full articles was conducted. Two physicians or a physician and a health services researcher reviewed each article. Articles were not masked prior to review. Discrepancies on inclusion/exclusion were resolved between the reviewers.

Table 12. Reason for Rejection of Full-Length Articles (more...)

Table 12. Reason for Rejection of Full-Length Articles at Secondary Review

Reason Code	Reason for Rejection	Cochrane and Medline N=614	AAP Data Files N=29	Symposia on OME N=107	Technical Experts N=17	Articles and books N=31	Total N=798a
R0	Non-English language	1	0	0	0	0	1 (0.1%)
R1	Case report/editorial/ letter/clinical/ overview/ practice guidelines/ consensus statements	142	24	22	4	17	209 (26.2%)
R2	Non-human subjects	4	0	0	0	2	6 (0.8%)
R3	Study condition not OM	33	0	0	2	2	37 (4.6%)
R4	Age of study population >22 years	0	0	2	0	1	3 (0.4%)
R5	Study population exclusively one of the following: craniofacial defects, primary mucosal disorders, immunodeficiencies, or Down or other genetic syndromes	3	0	1	2	0	6 (0.8%)
R7	Key questions not addressed	354	5	74	9	2	444 (55.6%)
R8	Duplicate citation	39	0	3	0	7	49 (6.1%)
R9	Data not abstractable from article	38	0	5	0	0	43 (5.4%)

a

3 incorrect citations not included.

Table 13. Results of Secondary Screening of Full-Length (more...)

Table 13. Results of Secondary Screening of Full-Length Articles

	Cochrane and Medline	AAP Data Files	SymposiaProceedings on OME	Technical Experts	Articles and books	All Sources
Total Citations	3200	1441	159a	39	40	4879
Number of articles reviewed at secondary screening	975b	32	159	36	40	1242
Number accepted after secondary screening	372	3	52	19	5	451
Question 1	127	0	14	0	0	141
Question 2	74	1	20	17	0	112
Question 3	157	2	22	2	3	186
Question 4	69	0	4	0	2	75
Total of above c	427	3	60	19	5	514

a

Exact number of citations not determined.

b

Four cases previously rejected because of age limit >12 were added to the original 971 accepted citations.

c

The ‘total of above’ number can exceed the number accepted because an article can address more than one question.

Of the 1,250 full length articles to be retrieved, 3 were irretrievable due to incorrect citation information. Secondary review of the remaining 1,247 full length articles from the various sources resulted in the rejection of 798 articles and acceptance of 449 articles: 141 for question 1, 112 for question 2, 186 for question 3 and 75 for question 4. Table 12 provides the reasons for rejection of the 798 articles and Table 13 summarizes the number of articles accepted during the secondary review process. During the fourth conference call, the experts raised the age limit to 22 for the responses to questions 2 and 3 to allow for detection of speech, language, and hearing problems past age 12, the original upper age limit. As a result, we revisited all titles/abstracts and articles that had been rejected because of the age limit, and four previously rejected articles were accepted from our original databases.

After establishing the analytical plan and before data abstraction, a physician and one health services researcher carried out a tertiary review of the 449 articles according to a set of established criteria for each key question. During this tertiary review, study design and quality were also evaluated.

For Question 1, we used three criteria for tertiary screening: 1) was the study a prospective cohort(s) study or a randomized control trial (RCT), 2) whether the control group in the RCT used the other ear as the ‘unit of control’ or not, and 3) were the outcome data abstractable?

For Questions 2 and 3, we established 5 criteria for tertiary screening:

1. Was the degree of OME determined for the first 3 years of life, and could the OM degree for the period before 3 years of age be linked to a specific outcome? If a study began at age 3, the study was not considered to fulfill this criterion.

2. Was the upper age limit 22 years of age? If a study included subjects older than 22 years, this criterion could be fulfilled only if outcomes for the 22 years of age and under was reported.

3. Was the degree of OM graded in some way, such as total time with OM, some measure of OM persistence, OM recurrence, or some measure of OM severity, and could the OM degree grade be linked to a specific outcome?

4. Was the study prospective? A study is considered prospective if the outcomes were measured prospectively. Cross-sectional and case-control studies were specifically excluded. A study that followed subjects prospectively for both OM history and outcome measures was considered a prospective cohort study. Studies that followed subjects prospectively for outcome measures (i.e. over a period of time) but retrospectively for OM history were considered to be retrospective-prospective studies and were accepted for inclusion. Studies that collected data on the outcome measures at one point in time, were considered retrospective cohort studies regardless of whether OM history was collected prospectively or retrospectively and were excluded. Studies that presented a cross-sectional analysis of prospectively collected data were not considered prospective and were excluded. Randomized controlled studies of an intervention with longitudinally measured outcomes were considered prospective and were included.

5. Was the outcome measured when the child was older than 3 years of age?

For Question 4, we established 3 criteria for inclusion during tertiary review: 1) Were the diagnostic procedure of interest and the reference standard performed within 24 hours of each other? 2) Is the diagnostic procedure not an algorithm or combination of multiple diagnostic procedures, 3) Are the reference standards one of those specified in the scope (tympanocentesis, MRI, myringotomy, validated pneumatic otoscopy, or CT scan), and 4) Are the data abstractable.

After tertiary review and data abstraction, a total of 114 articles were included in this evidence report, five of which addressed more than one key question:

• 38 chort studies for question 1, the natural history question;

• 21 studies for question 2, the speech and language question;

• 8 studies for question 3, the hearing question; and

• 52 studies for question 4, the diagnostic method question.

Table 14. Results of Tertiary Screening of Full-Length (more...)

Table 14. Results of Tertiary Screening of Full-Length Articles for Analysis

	Accepted at Secondary Screening	Abstracted in Evidence Tables
	449	114 (25%)
Question 1	141	38 (27%)
Question 2	112	21 (19%)
Question 3	186	8 ( 4%)
Question 4	75	52 (69%)
Total of above a	514	119 (23%)

a

An article can address more than one question.

Table 14 presents the results of tertiary review of the 449 articles.

Review and Assessment of Study Quality

We established criteria used for the assessment of study quality prior to the review of articles. Only prospective cohort studies were reviewed for Questions 1, 2 and 3 because of concerns about the validity of case-control, cross-sectional, and retrospective cohort studies. Diagnostic studies were reviewed for Question 4. The criteria used to evaluate the quality of both types of studies were modified from the work by the McMaster University Group (Jaeschke, Guyatt, and Sackett, 1994; Sackett, 1981; Trout, 1981; Tugwell, 1981).

The quality of natural history studies was evaluated against the following criteria:

1. Was the study a prospective cohort study?

2. Was the outcome(s) of the study clearly defined?

3. Was the out come(s) measured at a clearlly defined timepoint(s)?

4. Was the cohort of subjects followed without any intervention?

5. Was there blinded assessment of the outcome(s) of the study?

6. Were point estimates and measures of variability provided for the main adverse outcomes measured?

The quality of prospective cohort studies was evaluated against eight components:

1. Was the study cohort(s) clearly defined, with clearly spelled out inclusion and exclusion criteria?

2. Was the study cohort(s) assembled at a uniform point in the course of the child's illness?

3. Were the pathways by which patients entered the study clearly described?

4. Was complete follow-up achieved?

5. Were withdrawals and drop-outs described?

6. Were objective outcome criteria developed and used?

7. Was the outcome assessment “blind”?

8. Was adjustment for extraneous factors carried out?

The quality of diagnostic studies was evaluated against six components:

1. Was the reference standard appropriate?

2. Were the test results and the reference standard assessed independently of each other?

3. Were the readers of the results of the diagnostic test or the reference standard blinded?

4. Did the patient sample include an appropriate spectrum of mild and severe, treated and untreated patients to whom the diagnostic tests were applied in clinical practice?

5. Were the reproducibility of the test result (precision) and its interpretation (observer variation) determined?

6. Were the methods for performing the test described in sufficient detail to permit replication?

Articles were not masked prior to review. The Task Order Coordinator resolved minor discrepancies between the two reviews of each article. Conferences were held to resolve discrepancies whenever needed.

Data Abstraction

For the articles deemed eligible for inclusion in the Evidence Report, data abstraction was carried out by a two-member team that consisted of a physician reviewer and a health services researcher. One of the two members abstracted the data onto the evidence table, and the other member checked the data for accuracy. Data abstracted included parameters necessary to define study groups, inclusion/exclusion criteria, influencing factors, and outcome measures to be used in analysis.

Specific instructions for data abstraction were recorded. For Question 1, the outcome indicators for abstraction included partial OME resolution (resolution in one ear for bilateral OME only), complete OME resolution, relapse/recurrence (fluctuation/dynamic course), AOM after OME. For Question 2, the outcome indicators for abstraction were expressive or receptive language, expressive or receptive speech, and cognitive verbal intelligence. For Question 3, the outcome indicators for abstraction included conductive or sensorineural hearing loss. For Question 4, the outcome indicators for abstraction were sensitivity, specificity, positive and negative predictive values, and prevalence rate. For all questions, the time or age at which each outcome was measured was recorded. The outcome measures included both continuous and categorical measures. Continuous measurements included mean time or a median time and the categorical measures included proportion with resolution at specified times where both numerators and denominators were recorded. A key issue was that individual children, not populations, must be tracked. The latter would be acceptable only if we knew for certain that the same children were checked at both times.

Procedures to Reduce Bias, Enhance Consistency, and Check Accuracy

To reduce selection bias, we assigned two physician reviewers to screen and review titles/abstracts and full articles at every stage of the selection process. We assigned one physician and one health services researcher who was familiar with experimental design and biostatistics to abstract data. We assessed completeness of our collection of retrieved articles by cross-checking with studies included in other meta-analyses and references listed in review articles. The software program EndNote was used to check batches of articles added to the master list for duplicate references by comparing author, year title, and reference type. Following the importation of the first literature search, we used the software program EndNote to check subsequent references for duplication prior to their addition to the master list. After the master list was completed, we performed a second, manual, check to ensure no duplication.

To assess the extent of publication bias, we searched multiple sources and unpublished material identified by the Technical Expert Panel and internal content experts. We also studied funnel plots—scatter plots of sample size versus the estimated effect size from each study. When publication bias existed, a portion of points would be missing from the funnel plot, typically at the null effect level. Because graphical evaluation can be subjective, we also conducted an adjusted rank correlation test (Begg, 1999) and a regression asymmetry test (Egger, Smith, Schneider et al., 1997) as formal statistical tests for publication bias. We conducted these tests using the statistical package Stata (StataCorp. 1999).

The mechanisms used to enhance consistency in screening and data abstraction include the use of pre-designed forms with explicit instructions and continuous and prompt resolution of discrepancies. Data were entered into a Microsoft Excel spreadsheet directly by the screeners or data abstractors. A third project staff cross-checked data for individual studies abstracted by each data collector. We resolved discrepancies by rechecking the article or by consensus via conference calls.

Preparation of Evidence Tables

An evidence table was prepared for each key question. Each evidence table provides a comprehensive tabular display of data abstracted from the literature in response to the question. It contains the name of the first author, year of publication, study design and quality score, how and by whom OME diagnosis was done, when and where the study took place, inclusion and exclusion criteria, important influencing factors, sample size, outcome measures and their definitions, and study findings. A total of four evidence tables was prepared; they are included in the section called Evidence Tables in this report.

Supplemental Analysis

Based on the discussions of technical experts during the conference calls and the designated time frame for the evidence assessment, a supplemental analysis plan was developed for each key question to synthesize the data.

Identification of Peer Reviewers

Table 15. Peer Review Panel

Table 15. Peer Review Panel

Peer Reviewer	Area of Expertise	Affiliation/Location
Howard Bauchner, M.D.	General Pediatrics	Child and Adolescent Health Scholar in Residence Agency for Healthcare Research and Quality
Hanan S. Bell, Ph.D.	Methodology reviewer	Seattle, WA
Alfred O. Berg, MD, MPH	Family Medicine	University of Washington, Seattle, WA
Patricia A. Fall, MS, CRNP	Nurse Practitioner	Wexford, PA
George A. Gates, MD	Otolaryngology	University of Washington, Seattle, WA
Janice Goertz, RN, CPNP	Nurse Practitioner	Portage, MI
Judith Gravel, PhD	Hearing and Speech	Albert Einstein College of Medicine, Bronx, NY
Mark P.Haggard, Ph.D.	Hearing/ Psychoacoustics	Institute for Hearing Research, Nottingham, UK
Vic Hasselblad, Ph.D.	Meta-analysis reviewer	Duke University, Durham, NC
Tracy Lieu, MD	Pediatrician/Health Plan	Harvard Pilgrim Health Care, Boston, MA
Martin C. Mahoney, MD, Ph.D.	Family Medicine	DeGraff Family Medicine, North Tonawanda, NY
A. Richard Maw MS FRCS	Otolaryngology	Bristol Royal Infirmary, Bristol UK
Robert Ruben, MD	Otolaryngology	Montefiore Medical Center, Bronx, NY
Anne GM Schilder, MD, Ph.D.	Otolaryngology	University Medical Center Utrecht, The Netherlands
Steve Shelov, MD	Pediatrics	Scarsdale, N.Y
Sylvan Stool, MD	Otolaryngology	The Children's Hospital, Denver, CO
Robin Yurk, MD, MPH	Consumer/Health Plan	Community Clinic, Inc. Rockville, MD
Dr. J.O.M. Zaat (Joost Zaat)	Methodology reviewer	Purmerend, The Netherlands

Peer Review Process

Table 16. Instructions for Reviewing Draft Evidence (more...)

Table 16. Instructions for Reviewing Draft Evidence Report

Enclosed is a draft evidence report on the diagnosis and treatment of otitis media with effusion. You may make your comments either directly on the draft evidence report, or on a separate sheet of paper. If you choose to record your comments on a separate piece of paper, please use the page and paragraph number to identify to which part of the report your comments pertain.

We ask that you consider the following questions while you read this report. We realized that some of the questions may not pertain to your area of expertise. Please feel free to comment only on those that you feel most suited to answer.

Overall evaluation Is it clear what we did? You may agree or disagree with our methods, findings, or conclusions, but you should be able to understand what we did in order to produce this report. Methodology Are the methods we used appropriate: for identifying the key questions of interest from the panel of technical experts? for searching and reviewing the identified literature? for synthesizing the literature? Evidence Did we miss any crucial pieces of information in our literature search? Does the evidence support the conclusions? Utility Would you find this information to be useful if you had to develop clinical practice guidelines or medical review criteria for diagnosis and treatment of otitis media with effusion in children?

Key question 1: What is the Natural History (Spontaneous Resolution Rate over Time without Treatment) for

• OME persisting after a discrete episode of acute otitis media,

• newly diagnosed OME of unknown duration (unilateral or bilateral),

• OME persisting for weeks or months (unilateral or bilateral),

• unilateral OME lasting 3 months or longer,

• bilateral OME lasting 3 months or longer?

Literature Review

Table 17. Cohorts and Articles Relevant to Question (more...)

Table 17. Cohorts and Articles Relevant to Question 1 (n=40)

Cohort Identifier	Relevant Articles (ID, authors, year)	Comments
Birch	862 Birch and Elbrønd (1984)	0.75- to 7-year old children followed from 1/1982 to 4/1982.
Casselbrant I	1000 Casselbrant, Brostoff, Ashoff, and Bluestone (1985)	2- to 5-year old children followed from 9/1981 to 8/1983.
	2929 Casselbrant, Brostoff, Ashoff, and Bluestone (1990)
Casselbrant II	2929 Casselbrant, Brostoff, Ashoff, and Bluestone (1990)	5- to 12-year old children followed from 9/1984 to 5/1985.
Ernston	1202 Erston and Sundberg (1984)	Children 1- to 11-years old embedded in a controlled trial.
Fiellau-Nikolajsen I	1237 Fiellau-Nikolajsen and Lous (1979)	Children in Hjoerring, Denmark first examined in 1976.
	1235 Fiellau-Nikolajsen (1979)
	1242 Fiellau-Nikolajsen (1981)
	3051 Lous and Fiellau-Nikolajsen (1988)a
Fiellau-Nikolajsen II	1235 Fiellau-Nikolajsen (1979)	Children in Hjoerring, Denmark first examined in 1978.
	1245 Fiellau-Nikolajsen (1983)
Fiellau-Nikolajsen III	1777 Lous and Fiellau-Nikolajsen (1981)	Children in Hirtshals and Sindal, Denmark first examined in 1978.
Holmquist	1494 Holmquist, Fadala, and Qattan (1987)	7- to 9.5-year old children followed 2/1983 to 4/1983.
Lamothe	1714 Lamothe, Boudreault, Blancette, Tetreault, and Poliquin (1981)	First graders followed over a 6 week period in 1979.
Leiberman	1735 Leiberman and Bartal (1986)	2- to 12-year old children who had a follow-up exam after a 2.5 year delay in ventilating tube placement.
Marchisio	9 Marchisio, Principi, Salpietro, Boschi, Chetri, Caramia, Longhi, Reali, Meloni, DeSantis, Sacher, and Cupido (1998)	Primary school children followed for 12 weeks after the initial exam, and then a subset were randomized to a placebo group for another 8 weeks of follow-up.
Mills	1927 Mills and Vaughan-Jone (1992)	Prospective single cohort embedded in a comparative cohort of children 1- to 14-years old who had a follow-up exam about 2 months after the initial visit.
Portoian-Shuhaiber	2184 Portoian-Shuhaiber and Cullinan (1984)	5- to 6-year old children followed for 10 weeks in 1979.
Renvall I	2240 Renvall, Lidén, Jungert, and Nilsson (1978)	10- to 11- year old children examined after a 3-year interval.
Renvall II	2242 Renvall, Anniansson, and Lidén (1982)	4-year old children followed over a 12 week period in 1980.
Reves	2243 Reves, Budgett, Miller, Wadsworth, and Haines (1985)	3- to 6-year old children followed 11/1983 to 2/1984.
Roberts	2262 Roberts, Johnson, Carlin, Turczyk, Karnutta, and Yaffee (1995)	Newborns followed for 2 months after birth.
Robinson	2270 Robinson, Allen, and Root (1988)	Infants followed for 6 weeks.
Sly I	2457 Sly, Zambie, Fernandes, and Frazer (1980)	4- to 5-year old children recruited in 2/1977.
Sly II	2457 Sly, Zambie, Fernandes, and Frazer (1980)	4- to 5-year old children recruited in 9/1977.
Tos I	1486 Holm-Jensen, Sørenson, and Tos (1981)	Children born in 1975 and followed from 5/1979 to 2/1985.
	2636 Tos (1981)a
	543 Tos, Holm-Jensen, Sørenson, and Mogensen (1982)
	2639 Tos (1983)a
	2642 Tos (1984)a
	4834 Tos (1984)a
	4835 Tos (1988)a
Tos II	2629 Tos (1979)	Children born in 1976 and followed from 11/1977 to 2/1985.
	2190 Poulsen and Tos (1980)
	2631 Tos (1980)a
	2634 Tos (1980)
	2593 Thomsen and Tos (1981)
	2639 Tos (1983)a
	2642 Tos (1984)a
	4834 Tos (1984)a
	4835 Tos (1988)a
Tos III	2189 Poulsen and Tos (1978)	Children born in 1977 and followed from 1–2/1977 to 2/1985.
	2627 Tos (1979)
	2631 Tos (1980)a
	2634 Tos (1980)
	2639 Tos (1983)a
	2642 Tos (1984)a
	4834 Tos (1984)a
	4835 Tos (1988)a
Tos IV	1946 Moller and Tos (1990)	Children checked daily for 30 days.
van Balen	91 van Balen, De Melker, Touw-Otten (1996)	6-month to 6-year old children followed for 3 months in the early 1990's.
Williamson	2791 Williamson (1994)	5- to 8-year old children followed from 1988-1989 to 1991.
Wilmot	2795 Wilmot (1988)	6-month to 10-year old children were followed for 12 months after developing OME after AOM.
Zielhuis	2863 Zielhuis, Rach, and van den Broek (1990)	2- to 4-years old children followed from 1982-1983.

a

The article did not include abstractable data relevant to the specific Question 1 outcome measures.

Table 18. Disposition of Articles From Rosenfeld's (more...)

Table 18. Disposition of Articles From Rosenfeld's Natural History Assessment

ID Number	Author	Year	Included	Excluded	Reason for Exclusion From Assessment
1000	Casselbrant	1985	X
1238, 1240	Fiellau-Nikolajsen	1980		X	1238 is a study of tympanometry as a diagnostic tool. 1240 is the same cohort as in 1245 and does not present sufficient data tracking individual cases or episodes of OME.
2242	Renvall	1982	X
2593	Thomsen	1981	X		2593 uses same cohort as 2190, 2629, 2631, 2639, 2642, 4834, and 4835.
2627	Tos	1979	X
2634	Tos	1980	X
543	Tos	1982	X		543 uses same cohort as 1486, 2636, 2639, 2642, 4834, and 4835.
2791	Williamson	1994	X
2863	Zielhuis	1990	X
2857	Zeisel	1995		X	2857 included 13% with purulent OME. Antibiotic administration and other interventions for either purulent or non-purulent OME. Insufficient data track individual cases or episodes of OME.
2243	Reves	1985	X
91	van Balen	1996	X		Randomized controlled trial with an initial 3-month watchful waiting period of children with OME.
1777	Lous	1981	X

Table 19. Question 1: Articles Excluded During Data (more...)

Table 19. Question 1: Articles Excluded During Data Abstraction

ID#	Author	Year	Reason Not Included
705	Aniansson	1985	Screening study over 2 years. Some retest of same subjects, but no control over treatment.
3051	Lous	1988	Data not abstractable as presented. Available counts are in essential agreement with articles 1235, 1237, and 1242 apart from minor differences in number of tympanogram types, e.g. type A in Jan 1975 (629 in articles 1235, 1237, and 3051 and 631 in article 1242) and types B, C1, and C2 in Jun-July 1976 (32, 31,, and 42 respectively in articles 1235, 1242, and 3051 and 37, 28, and 40 respectively in article 1237).
2578	Teele	1980	Data not abstractable; data on persistent OME after AOM are not presented and cannot be derived.
2631	Tos	1980	Article does not present any new information other than the actual initial counts of tympanogram types for the 1976 and 1977 Tos cohorts utilized in the evidence tables for article 2629 and 2627.
2636	Tos	1981	Data not abstractable as presented relative to the data on 3- and 6-month follow-up presented in article 1486. Individual cases of type B, C1, or C2 tympanograms cannot be tracked from 2/1979 to 11/1979 or 2/1980.
2639	Tos	1983	The only relevant new information is that the authors mention that in the 1976 cohort, 50% of tympanogram type B changed to types A or C over the first 3 months of the study. However, the number with type B who presented for the 3-month follow-up exam is not given so per cent change cannot be calculated.
2642	Tos	1984	Article does not present any abstractable data relevant to Q1.
4834	Tos	1984	Article does not present any new data relevant to Q1.
4835	Tos	1988	Article does not present any new data relevant to Q1.
2795	Wilmot	1988	Article presented data on OME following AOM which was eliminated as a condition of interest because it had been studied in a recent evidence analysis.

After initial screening of 4,879 titles or abstracts, we identified 449 articles for review. After secondary screening of the 449 articles, we identified 141 articles that fell within the scope of this question. Tertiary screening identified 38 articles on prospective cohort studies for potential abstraction. After reassessing articles included in the systematic review by Rosenfeld (1999), we included three more articles for potential abstraction. We eliminated one article on r OME following acute otitis media which was addressed in a recent evidence analysis. A total of 26 prospective cohort studies and one retrospective-prospective cohort study were identified among these 40 articles. Table 17 lists the studies and cohorts examined. Abstraction was possible from 33 of the 40 articles. Table 18 lists the studies referred to by Rosenfeld (1999) and their disposition in the present evidence-based analysis. Table 19 lists the articles excluded because relevant data could not be abstracted.

Findings

Table 20. Study Quality for Studies Included in Evidence (more...)

Table 20. Study Quality for Studies Included in Evidence Table on Natural History

ID	Author	Year	Study Quality Scorea
862	Birch	1984	3 (1,1,1,0,0,0)
1000	Casselbrant	1985	3 (1,1,1,0,0,0)
2929	Casselbrant	1990	3 (1,1,1,0,0,0)
1202	Ernstson	1984	2 (1,0,0,1,0,0)
1235	Fiellau-Nikolajsen	1979	4 (1,1,1,0,0,1)
1237	Fiellau-Nikolajsen	1979	4 (1,1,1,0,0,1)
1242	Fiellau-Nikolajsen	1981	4 (1,1,1,0,0,1)
1245	Fiellau-Nikolajsen	1983	3 (1,1,1,0,0,0)
1486	Holm-Jensen	1981	3 (1,1,1,0,0,0)
1494	Holmquist	1987	3 (1,1,1,0,0,0)
1714	Lamothe	1981	4 (1,1,1,1,0,0)
1735	Leiberman	1986	2 (1,1,0,0,0,0)
1777	Lous	1981	3 (1,1,1,0,0,0)
9	Marchisio	1998	3 (1,1,1,0,0,0)
1927	Mills	1992	1 (1,0,0,0,0,0)
1946	Moller	1990	3 (1,1,1,0,0,0)
2184	Portoian-Shuhaiber	1984	3 (1,1,1,0,0,0)
2189	Poulsen	1978	3 (1,1,1,0,0,0)
2190	Poulsen	1980	3 (1,1,1,0,0,0)
2240	Renvall	1978	3 (1,1,1,0,0,0)
2242	Renvall	1952	4 (1,1,1,1,0,0)
2243	Reves	1985	3 (1,1,1,0,0,0)
2262	Roberts	1995	4 (1,1,1,0,1,0)
2270	Robinson	1988	1 (1,1,0,0,0,0)
2457	Sly	1980	3 (1,1,1,0,0,0)
2593	Thomsen	1981	3 (1,1,1,0,0,0)
2627	Tos	1979	3 (1,1,1,0,0,0)
2629	Tos	1979	3 (1,1,1,0,0,0)
2634	Tos	1980	3 (1,1,1,0,0,0)
543	Tos	1982	3 (1,1,1,0,0,0)
91	van Balen	1996	3 (1,1,1,0,0,0)
2791	Williamson	1994	2 (1,1,0,0,0,0)
2863	Zielhuis	1990	3 (1,1,1,0,0,0)

a

The six components of study quality are: a prospective cohort study; outcome clearly defined; time point at which outcome measured clearly defined; subjects followed without any intervention; blinded assessment of outcome; and point and variability estimates provided for main outcome measures. 1 indicates presence and 0 indicates absence

Evidence Table 1 presents the study characteristics, population characteristics, risk factors, and findings for the 27 cohort studies described in 34 articles that responded to this question. Table 20 presents the study quality scores for the studies included in Evidence Table 1. The quality scores (see Methods) for these studies fell at the low end of the possible range of 1(lowest) -to 6(highest). Three of the cohort studies had a score of 4, sixteen had a score of 3, seven had a score of 2, and one had a score of 1.

Table 21. Tympanometry Definitions in Natural History (more...)

Table 21. Tympanometry Definitions in Natural History Cohorts Utilizing Tympanometry as the Sole Diagnostic Method

Tympanogram Type	A		As		B
Cohorta	Pressure (mmH2O)	immitance	Pressure (mmH2O)	immitance	Pressure (mmH2O)	immitance	Comments
Birch	> -100					< 0.25mlc, 1	1stapedial reflex absent and max compliance unreadable; Madsen Electronics tympanoscope, model ZS 330, 226 Hz probe tone
Fiellau-Nikolajsen I	> -100	> 0.1 d			200 to -4001	≤ 0.1d	1or indeterminable
Fiellau-Nikolajsen II							middle-ear effusion=flat curve or <= -100 with absent middle ear reflexes
Fiellau-Nikolajsen III	> -100					multiple criteria1	1type B=otoadmittance < 0.20millimhos, absolute gradient < 0.04millimhos and absence of ipsilateral acoustic reflex; Grason-Stadler Middle Ear Analyzer 1722
Holmquist	50 to -99				flat curve		Madsen ZA 330, 226 Hz probe tone
Portoian-Shuhaiber							abnormal defined as an abnormal tympanometric curve and/or absent acoustic reflex; Grason-Stadler Middle-ear Analyser (Model 1722)
Tos I	0 to -99				flat curve		Madsen ZO 70 tympanometer, 220 Hz probe tone
Tos II	0 to -99				flat curve		Madsen ZO 70 tympanometer, 220 Hz probe tone
Tos III	>-100				flat curve1		1<= 0.1e; Madsen ZO 70 impedance meter
Tos IV AZ7	99 to -99				flat curve without impedance minimum1		1or with a measurable impedance minimum and relative gradient < 0.1; Impedance audiometer AZ 7 (Interacoustics)
ZS 331	99 to -99				flat training1		1or compliance below 0.25mlc and absent ipsilateral stapedial reflex; Impedance tympanoscope ZS 331 (Madsen Electronics)
Reves	-100 to 50	>0.3b			< -100	lowb	tympanometer 85 AR 11 (American Electro Medics)
Robinson	-149 to +50	>0.2mlc					types As, B, C, and Cs are failures; Maico MA 610 portable impedance screener, 226 Hz probe tone
Sly I and II					flat curve1		1or compliance < 0.3mlc or peak compliance occurred at or below -100 mmH20; Teledyne Avionics acoustic impedance meter model TA-1D
van Balen	-99 to 200	≥ 0.2mmhoc			≤ -400	< 0.2mmhoc
Williamson	200 to -99				flat curve		Grayson-Stadler [sic] Earscan impedance audiometer
Zielhuis	≥ -99	≥ 0.2mlc			≤ -400	< 0.2mlc	Grason-Stadler-model 27

a

see Table 17 for associated cohort articles

b

impedance

c

admittance

d

relative gradient

e

impedance slope

Table 21a. Tympanogram Definitions in Natural History (more...)

Table 21a. Tympanogram Definitions in Natural History Cohorts Utilizing Tympanometry as the Sole Diagnostic Method

Tympanogram Type	C		C1		C2
Cohorta	Pressure (mmH2O)	immitance	Pressure (mmH2O)	immitance	Pressure (mmH2O)	immitance	Comments
Birch	≤ -100
Fiellau-Nikolajsen I			-100 to -199	>0.1d	-200 to -400	>0.1d
Fiellau-Nikolajsen II							middle-ear effusion=flat curve or <= -100 with absent middle ear reflexes
Fiellau-Nikolajsen III	≤ -100
Holmquist	100 to -300
Portoian-Shuhaiber							abnormal defined as an abnormal tympanometric curve and/or absent acoustic reflex
Tos I			-100 to -199		-200 to -350
Tos II			-100 to -199		-200 to -350
Tos III	-100 to -300	>0.1e	-100 to -199		-200 to -350
Tos IV AZ 7			-100 to -199		>-200
ZS 331			-100 to -199		>-200
Reves	< -100	>0.3b
Robinson							types As, B, C, and Cs are failures
Sly I and II
van Balen			-199 to -100	≥ 0.2mmhoc	-399 to -200	≥ 0.2mmhoc
Williamson			-100 to -199		-200 to -400
Zielhuis			-100 to -199	≥ 0.2mlc	-200 to -399	≥ 0.2mlc

a

see Table 17 for associated cohort articles

b

impedance

c

admittance

d

relative gradient

e

impedance slope

Although we accepted these 27 studies for the natural history analysis, half of them (thirteen) failed to document that the subjects had received no medical or surgical treatment during the course of the study that could affect the outcome of OME. In the three studies that claimed that their subjects received no treatment, the investigators failed to document how adherence was maintained or confirmed. In studies of eleven cohorts, the authors mentioned that children received antibiotics or underwent surgical procedures that could affect OME outcome. In those studies that reported numbers of children who received treatment, the proportion was relatively small compared to the total number studied. Except in two studies, results were not stratified by treatment condition, even in those articles that reported treatments received by study subjects. Further, the majority of studies, but not all, used tympanometry as the sole diagnostic test of OME. However, the criteria for each tympanogram type in these studies are similar (Table 21). Table 21 lists the pressure and immittance parameters for tympanogram types for those studies that used tympanometry as the sole diagnostic criteria for OME.

Table 22. OME Resolution by Ears on Newly Diagnosed (more...)

Table 22. OME Resolution by Ears on Newly Diagnosed OME of Unknown Duration, <6-month old cohorts

				Ears Resolved Intervala
Cohort ID	Diagnostic Methodb	antibioticc	surgeryc	<2wk	<2m	<3m	<6m	Article(s)
<6-month old cohorts
Robertse	oto	unknown	unknown	22/24 (92%)	24/24 (100%)			2262 Roberts 1995
Tos III	Tymp	unknown	unknown			1/4d (25%)	1/4 d (25%)	2627 Tos 1979

a

interval calculated from cohort inception and not cumulative, unless otherwise noted

b

oto=otoscopy, tymp=tympanometry (type B to A transition)

c

Did any of the patients receive antibiotic or surgery?

d

interval started at 6-month follow-up

e

cumulative resolution rate

Table 22a. OME Resolution by Ears on Newly Diagnosed (more...)

Table 22a. OME Resolution by Ears on Newly Diagnosed OME of Unknown Duration, 6 month to 3 year old cohorts

				Ears Resolved Intervala
Cohort ID	Diagnostic Methodb	antibioticc	surgeryc	<6wk	<3m	<6m	<9m	<24m	Article(s)
6-month to 3-year old cohorts
Robinson	tymp	unknown	unknown	10 of 25 d (40.0%)					2270 Robinson 1988
									1 2634 Tos 1980;
									2 2190 Poulson 1980;
					6/51 1	15/59 2	19/51 3	9/48 4	3 2629 Tos 1979;
Tos II	tymp	unknown	yes		(12%)	(25%)	(37%)	(19%)	4 2593 Thomsen 1981
Tos IIee	tymp	unknown	yes		6/51 1 (12%)	16/511 (31%)	24/511 (47%)		1 2634 Tos 1980

a

interval calculated from cohort inception and not cumulative, unless otherwise noted

b

oto=otoscopy, tymp=tympanometry (type B to A transition)

c

Did any of the patients receive antibiotic or surgery?

d

interval is minimum of 6wk so may be greater

e

cumulative resolution rate

Table 22b. OME Resolution by Ears on Newly Diagnosed (more...)

Table 22b. OME Resolution by Ears on Newly Diagnosed OME of Unknown Duration, > 3 year old cohorts

				Ears Resolved Intervala
Cohort ID	Diagnostic Methodb	antibioticc	surgeryc	<2wk	<3wk	<1m	<6wk	<3m	<4m	<6m	<8m	<1y	<3y	Article(s)
>3-year old cohorts
Fiellau-Nikolajsen Id	tymp	unknown	yes			14/941 (15%)		22/911 (24%)		32/691 (46%)			33/652 (51%)	11237 Fiellau-Nikolajsen 1979; 21242 Fiellau-Nikolajsen 1981
Fiellau-Nikolajsen IId	tymp	unknown	unknown			7/64 (11%)		16/62 (26%)						1235 Fiellau-Nikolajsen 1979
Holmquist	tymp	unknown	unknown					251/511 (49%)						1494 Holmquist 1987
Lamothe	pneum oto	no	no		24/64 (38%)		25/53 (47%)							1714 Lamothe 1981
Lamothed	pneum oto	no	no		24/64 (38%)		38/53 (72%)							1714 Lamothe 1981
Renvall I	tymp	unknown	unknown										282/335 (84%)	2240 Renvall 1978
Renvall IId	tymp	no	no				10/40 (25%)	16/40 (40%)						2242 Renvall 1982
Sly	tymp	no	no	1/9 (11%)		4/9 (44%)	6/9 (67%)							2457 Sly 1980
Sly II	tymp	no	no	0/5 (0%)		0/5 (0%)	0/5 (0%)							2457 Sly 1980
Tos I	tymp	unknown	unknown					3/921 (3%); 3/872 (4%)		14/931 (15%); 14/872 (16%)				11486 Holm-Jensen 1981; 2543 Tos 1982
Tos Id	tymp	unknown	unknown					3/871 (3%)		17/871 (20%)				1543 Tos 1982
Williamson	tymp	unknown	yes						35/67 (52%)		52/67 (78%)	61/67 (91%)		2791 Williamson 1994

a

interval calculated from cohort inception and not cumulative, unless otherwise noted

b

oto=otoscopy, pneum oto=pneumatic otoscopy, tymp=tympanometry (type B to A transition)

c

Did any of the patients receive antibiotic or surgery?

d

cumulative resolution rates

Table 22c. OME Resolution by Ears on Newly Diagnosed (more...)

Table 22c. OME Resolution by Ears on Newly Diagnosed OME of Unknown Duration, Age not stratifiable

				Ears Resolved Intervala
Cohort ID	Diagnostic Methodb	antibioticc	surgeryc	<1m	<2m	<3m	<4m	<5m	<6m	<3y	Article(s)
Age not stratifiable
Casselbrant Id	algorithm	yes	yes	92/137 (67%)	109/137 (80%)	130/137 (95%)	134/137 (98%)	136/137 (99%)	137/137 (100%)		1000 Casselbrant 1985
Renvall I	tymp	unknown	unknown							282/335 (84%)	2240 Renvall 1978

a

interval calculated from cohort inception and not cumulative, unless otherwise noted

b

algorithm=algorithm based on pneumatic otoscopy, tympanometry, and acoustic reflex, tymp=tympanometry (type B to A transition)

c

Did any of the patients receive antibiotic or surgery?

d

cumulative resolution rate

Table 23. OME Resolution by Childon Newly Diagnosed (more...)

Table 23. OME Resolution by Childon Newly Diagnosed OME of Unknown Duration, >3 year old cohorts

				Resolved intervala
Cohort ID	Diagnostic methodb	antibioticc	surgeryc	<2wk	<1m	<6wk	<10wk	<3m	<4m	<6m	<8m	<1y	Article(s)
>3-year old cohorts
Fiellau-Nikolajsen II	tymp	unknown	unknown		28/81 (35%)			46/80 (58%)		53/78 (68%)			1245 Fiellau-Nikolajsen 1983
Marchisio	pneum oto, tymp	unknown	no					325/451 (72%)					9 Marchisio 1998
Portoian-Shuhaiber	tymp	unknown	unknown				65/130 (50%)						2184 Portoian-Shuhaiber 1984
Sly I	tymp	no	no	1/7 (14%)	3/7 (43%)	5/7 (71%)							2457 Sly 1980
Sly II	tymp	no	no	0/3 (0%)	0/3 (0%)	0/3 (0%)							2457 Sly 1980
Williamson	tymp	unknown	yes						22/50 (44%)		38/50 (76%)	45/50 90%)

a

interval calculated from cohort inception unless otherwise noted

b

oto=otoscopy, tymp=tympanometry (type B to A transition)

c

Did any of the patients receive antibiotic or surgery?

Table 23. OME Resolution by Childon Newly Diagnosed (more...)

Table 23. OME Resolution by Childon Newly Diagnosed OME of Unknown Duration, Age not stratifiable

				Resolved intervala
Cohort ID	Diagnostic Methodb	antibioticc	surgeryc	<2m	<3m	Article(s)
Age not stratifiable
Mills	pnem oto, tymp	unknown	unknown	57/192 (30%)		1927 Mills 1992
Reves	tymp	unknown	unknown		40/68 (59%)	2243 Reves 1985
van Balen	tymp	unknown	unknown		223/443 (50%)	91 van Balen 1996

a

interval calculated from cohort inception unless otherwise noted

b

oto=otoscopy, tymp=tympanometry (type B to A transition)

c

Did any of the patients receive antibiotic or surgery?

Tables 22 and 23 provide OME resolution rates by ear, as reported in the majority of the studies, and by child for those studies in which data could be stratified by unit of analysis (ear vs. child), age group, and OME type respectively. For clarity, we have listed only those resolution intervals with onset at the study's inception. For clarity, when tympanometry was the diagnostic method, counts for tympanogram type B transition to A are shown when provided, while other tympanogram transitions are shown in Evidence Table 1.

Data from a number of studies were reported in a format that made them unusable for quantitative syntheses. Ten studies were eliminated from quantitative syntheses, because they failed to stratify by age (i.e. less than and greater than 3 years old). Day-to-day variability in tympanogram types was described in a cohort of kindergarten children examined on each weekday for 30 days (Moller and Tos,1990). Ernston and Sundberg (1984) described a group of children who participated in a controlled trial of children with OME that persisted for at least 3 months; they found that 15.3 percent (11 of 72) of such children had OME resolution by five weeks followup. A study of children with OME that persisted for 3 months or more, showed 45 percent (49 of 109) OME resolution at 2.5 year followup (Leiberman and Bartal,1986). However, the investigators acknowledged that middle-ear effusion noted after such a long interval could be either persistent or recurrent. Birch and Elbrønd (1984) and Zielhuis, Rach, and van den Broek (1990) derived equations to describe the OME resolution rates they observed in their cohorts, but we were unable to abstract actual counts from these articles.

We performed two sets of meta-analyses that matched age groups, unit of analysis, outcome type, and time to resolution in three or more cohorts. All meta-analyses presented here used the ‘ear’ as the unit of analysis. Few studies considered the child or the episode as the unit of analysis and no meta-analyses were possible.

Table 24. Meta-Analysis for <6 Weeks Resolution (more...)

Table 24. Meta-Analysis for <6 Weeks Resolution Rate for Newly Diagnosed OME of Unknown Duration In Children Older Than 3 Years of Age

ArticleID	Author	Criterion	Age at diagnosis	Antibiotic used?	Surgery performed?	Followup interval	Number ears resolved	Total number ears	Resoluation rate in %	Random Effects Pooled Estimate (95% CI)	Test of Heterogeity Q statistic (P-value)
2457	Sly-1980	B or C to A	5yr	no	unknown	<6wk	18	32	56.3
2457	Sly-1980	B or C to A	5yr	no	unknown	<6wk	11	22	50.0
1714	Lamothe-1981	Otoscopy	6yr	no	unknown	<6wk	25	53	47.2
	Total						54	107	44.9	42.3 (24.1, 60.6)	7.85 (p=0.02)
ArticleID	Author	Criterion	Age at diagnosis	Antibiotic used?	Surgery performed?	Followup interval	Number ears resolved	Total number ears	Resoluation rate in %	Random Effects Pooled Estimate (95% CI)	Test of Heterogeity Q statistic (P-value)
2457	Sly-1980	B to A	5yr	no	unknown	<6wk	6	9	66.7
2457	Sly-1980	B to A	5yr	no	unknown	<6wk	0	5	0.0
1714	Lamothe-1981	Otoscopy	6yr	no	unknown	<6wk	25	53	47.2
	Total						31	67	46.3	37.2 (1.8, 72.5)	16.4 (p<0.001)

Note: Lamothe's study used otoscopy and is included in all meta-analyses

Resolution at 6-Week Followup The first set of meta-analyis contains two meta-analyses (Table 24). The meta-analyses showed that if the criteria for resolution were tympanogram type B or C transition to A or by otoscopy, 42.3 percent (95% CI: 24.1%, 60.6%) of ears with newly diagnosed OME of unknown duration in children older than 3 years had resolution by the 6-week exam. If the criteria for resolution were modified to tympanogram type B transition to A or otoscopy, the proportion of ears in children older than 3 years old with resolution at 6-week followup was 37.2 percent (95% CI: 1.8%, 72.5%). Spontaneous resolution rates were significantly different among the cohorts for both definitions. In these studies, the OME resolution rates were calculated by determining the proportion of children without OME at followup, whether or not their baseline OME had resolved and recurred at an earlier point. Thus, these are not cumulative resolution rates (Sly, Zambie, Fernandes et al., 1980; Lamothe, Boudreault, Blanchette et al., 1981).

Table 25. Meta-Analysis for <3 Months Cumulative (more...)

Table 25. Meta-Analysis for <3 Months Cumulative Resolution Rate for Newly Diagnosed OME of Unknown Duration In Children Older Than 3 Years of Age

ArticleID	Author	Criterion	Age at diagnosis	Antibiotic used?	Surgery performed?	Followup interval	Number ears resolved	Total number ears	Resoluation rate in %	Random Effects Pooled Estimate (95% CI)	Test of Heterogeity Q statistic (P-value)
1237	Fiellau-Nikolajsen-1979	B or C to A	3–4yr	unknown	unknown	<3mo	154	348	44.3
1235	Fiellau-Nikolajsen-1979	B or C to A	3–4yr	unknown	unknown	<3mo	83	200	41.5
543	Tos-1982	B or C to A	4yr	unknown	unknown	<3mo	103	393	26.2
2242	Renvall-1982	otoscopy	4yr	no	unknown	<12wk	86	144	59.7
	Total						426	1085	39.3	42.7 (29.3,56.1)	63.01 (p<0.001)
ArticleID	Author	Criterion	Age at diagnosis	Antibiotic used?	Surgery performed?	Followup interval	Number ears resolved	Total number ears	Resoluation rate in %	Random Effects Pooled Estimate (95% CI)	Test of Heterogeity Q statistic (P-value)
1237	Fiellau-Nikolajsen-1979	B to A	3–4yr	unknown	unknown	<3mo	22	91	24.2
1235	Fiellau-Nikolajsen-1979	B to A	3–4yr	unknown	unknown	<3mo	16	62	25.8
543	Tos-1982	B to A	4yr	unknown	unknown	<3mo	3	87	3.4
2242	Renvall-1982	otoscopy	4yr	no	unknown	<12wk	16	40	40.0
	Total						57	280	20.4	22.5 (5.9,39.0)	44.28 (p<0.001)

Resolution at 3-Month Followup The second set of meta-analyses are shown in Table 25. The first two meta-analyses assessed resolution of OME by ear by the time of 3-month followup in children older than 3 years. If the criteria for resolution were tympanogram type B or C transition to A, the proportion of ears with resolution of newly diagnosed OME of unknown duration was 42.7 percent (95% CI: 29.3%, 56.1%) among children older than 3 years. If the criteria for OME resolution were tympanogram type B transition to A, then the proportion of ears with resolution was 22.5 percent (95% CI: 5.9%, 39.0%). Spontaneous resolution rates were not significantly different among the studies in either comparison, except for Tos, Holm-Jensen, Sörenson, and Mogensen (1982) which had lower resolution rates. Because Fiellau-Nikolajsen and Lous (1979), Fiellau-Nikolajsen (1979), and Renvall, Anniansson, and Lidèn (1982) terminated followup whenever a child had a type A or normal tympanogram and because Tos, Holm-Jensen, Sörenson, and Mogensen (1982) provided data for calculation of the cumulative OME resolution rate, this estimate represents a cumulative resolution rate. The last three meta-analyses in Table 25 show the derivation of the cumulative resolution rates.

Resolution Rates for Younger Age Groups Only two studies examined the resolution rates for each of the age groups of less than 6 months and 3-months to 3-years (see Tables 22 and 23). Thus, no meta-analyses for children under 3 years of age were performed.

The Role of Influencing Factors in Resolution Similarly, because very few studies assessed the role of factors that might influence resolution, no meta-analyses were performed. The results of individual studies are summarized here:

• Lamothe, Boudreault, Blanchette, and colleagues (1981) and Fiellau-Nikolajsen and Lous (1979) assessed the effects of gender and found quicker resolution among females than males.

• Fiellau-Nikolajsen (1979) noted that children who received at-home care had quicker resolution of OME than did children in daycare.

• Sly, Zambie, Fernandes, and colleagues (1980) compared OME resolution in small cohorts first studied in February with those first studied in September and found resolution to be more rapid in the February cohort.

• Lamothe, Boudreault, Blanchette, and colleagues (1981) also assessed the effect of the side of the affected ear and noted more rapid resolution in affected right ears than affected left ears.

• Portoain-Shuhaiber and Cullinan (1984) stratified by racial/ethnic origin and noted quicker resolution of OME in African children than in Indian or Caucasian children.

• Moller and Tos (1990) found that different tympanometry instruments, which they described as impedance tympanoscopy and impedance audiometry, gave different rates of OME among the same group of children.

• Zielhuis, Rach, and van den Broek (1990) found that season and age at the end of the episode had a statistically significant effect on OME resolution rates, while gender, upper respiratory tract infection, and history of AOM did not, although they presented only percentages without denominators.

Key Question 2: What Are the Effects of Early-Life OM on Long-Term Speech and Language Development?

• Do infants and preschool children with longer-duration early-life OME as compared to those with shorter duration OME have greater delays in speech and language development (receptive or expressive) later in life?

• Is OME-associated conductive hearing loss in the first 3 years of life a risk factor for speech and language developmental delays?

• What are the risk factors that interact with the effect of OME on speech and language development in infants and preschool children?

Literature Review

Table 26. List of Cohort Studies Included for Question (more...)

Table 26. List of Cohort Studies Included for Question 2

ID#	Author	Year	Cohort
1623	Kaplan	1973	Eskimo villages in the Yukon and Kuskokwim River Delta areas of Southwestern Alaska
1255	Fischler	1985	Four Indian reservations in Arizona
4657	Roberts	1986	Frank Porter Graham Child Development Center, Chapel Hill, NC
3118	Roberts	1988	Frank Porter Graham Child Development Center, Chapel Hill, NC
4806	Roberts	1988	Frank Porter Graham Child Development Center, Chapel Hill, NC
4656	Roberts	1989	Frank Porter Graham Child Development Center, Chapel Hill, NC
3117	Roberts	1991	Frank Porter Graham Child Development Center, Chapel Hill, NC
4319	Roberts	1995	Frank Porter Graham Child Development Center, Chapel Hill, NC
1373	Gravel	1992	LIFE (Longitudinal Infant Follow-up and Evaluation) Program of the Rose F. Kennedy Center, Albert Einstein College of Medicine, Bronx, NY
4728	Gravel	1996	LIFE (Longitudinal Infant Follow-up and Evaluation) Program of the Rose F. Kennedy Center, Albert Einstein College of Medicine, Bronx, NY
1941	Mody	1999	LIFE (Longitudinal Infant Follow-up and Evaluation) Program of the Rose F. Kennedy Center, Albert Einstein College of Medicine, Bronx, NY
2295	Ruben	1997	LIFE (Longitudinal Infant Follow-up and Evaluation) Program of the Rose F. Kennedy Center, Albert Einstein College of Medicine, Bronx, NY
1219	Feagans	1987	Medical and Day Care Intervention Project in Pennsylvania
2135	Paul	1993	Portland Language Development Project (PLDP), Oregon
4651	Klein	1988	The Greater Boston Otitis Media Study Group, MA
2583	Teele	1990	The Greater Boston Otitis Media Study Group, MA
1435	Harsten	1993	University Hospital of Lund, Sweden
877	Black	1993	University of Maryland Medical System, Baltimore, MD
4675	Owen	1996	University of Texas Medical Branch, TX
1277	Freeark	1992	University-based pediatric clinic in Michigan

After secondary screening of the 449 articles retrieved for review, we identified 112 articles that fell within the scope of this question. Tertiary screening identified 20 studies that fulfilled the 5 criteria for analysis. (Table 26) The five criteria included the following:

1) OM was diagnosed/assessed before the age of 3 years, 2) speech or language outcome was measured at or before the age of 22 years, 3) a prospective cohort study design was used, 4) OM was graded, and 5) speech or language outcome was measured beyond 3 years of age. Of the 20 studies, 17 were prospective cohort studies and three (Freeark, 1992; Fischler, 1985; Paul, 1993) were retrospective-prospective studies in which OM history was retrospectively obtained, but outcome measures were prospectively obtained. We did not exclude any studies based on their quality.

When we reviewed the studies included in the 1994 OME Guideline for inclusion in our analysis (Stool, Berg, Berman et al., 1994), we excluded six of the studies. The Friel-Patti and Finitzo (1990) study was excluded, because it was not a prospective cohort study and did not report outcomes for children over 3 years of age. The Friel-Patti (1982) study was excluded because no outcomes were measured in children over 3 years of age. The studies of Lous and Fiellau-Nikolajsen (1988) and Rach, Ziehlhuis, van Baarle, and colleagues (1991 were excluded, because the investigators did not measure OM severity before the age of 3 years. The Rach, Zielhuis, and van den Broek (1988) study was excluded, because it was not a prospective cohort study. The study by Wright, Sell, McConnell, and colleagues (1988) was excluded, because no outcomes were measured in children over 3 years of age. Finally, unlike the OME Guideline (Stool, Berg, Berman et al., 1994), we included the studies by Klein (1988) and Freeark (1992), because study quality was not one of our rejection criteria.

Table 27. List of Cohort Studies Not Included for Question (more...)

Table 27. List of Cohort Studies Not Included for Question 2 (For Reason of Not Reporting Findings Beyond 3 Years of Age)

ID#	Author	Year	Cohort
3119	Roberts	1995	Frank Porter Graham Child Development Center, Chapel Hill, NC
4841	Wallace	1988	LIFE (Longitudinal Infant Follow-up and Evaluation) Program of the Rose F. Kennedy Center, Albert Einstein College of Medicine, Bronx, NY
2739	Wallace	1988	LIFE (Longitudinal Infant Follow-up and Evaluation) Program of the Rose F. Kennedy Center, Albert Einstein College of Medicine, Bronx, NY
2740	Wallace	1988	LIFE (Longitudinal Infant Follow-up and Evaluation) Program of the Rose F. Kennedy Center, Albert Einstein College of Medicine, Bronx, NY
4842	Wallace	1992	LIFE (Longitudinal Infant Follow-up and Evaluation) Program of the Rose F. Kennedy Center, Albert Einstein College of Medicine, Bronx, NY
667	Abraham	1996	LIFE (Longitudinal Infant Follow-up and Evaluation) Program of the Rose F. Kennedy Center, Albert Einstein College of Medicine, Bronx, NY
4796	Petinou	1996	LIFE (Longitudinal Infant Follow-up and Evaluation) Program of the Rose F. Kennedy Center, Albert Einstein College of Medicine, Bronx, NY
2742	Wallace	1996	LIFE (Longitudinal Infant Follow-up and Evaluation) Program of the Rose F. Kennedy Center, Albert Einstein College of Medicine, Bronx, NY
3096	Petinou	1999	LIFE (Longitudinal Infant Follow-up and Evaluation) Program of the Rose F. Kennedy Center, Albert Einstein College of Medicine, Bronx, NY
4671	Luloff	1993	Longitudinal study investigating the efficacy of drug prophylaxis on otitis media in greater Boston area, MA
4673	Tsushima	1993	Longitudinal study investigating the efficacy of drug prophylaxis on otitis media in greater Boston area, MA
4674	Wendler-Shaw	1993	Longitudinal study investigating the efficacy of drug prophylaxis on otitis media in greater Boston area, MA
875	Black	1988	Maryland Otitis Media Study Group, Baltimore
4708	Downs	1988	Not specified
2719	Vernon-Feagans	1996	Ongoing study of health and day-care in a semi-rural area of northeastern United States
1288	Friel-Patti	1982	Parkland Memorial Hospital, Dallas, TX
4713	Feldman	1996	Pittsburgh-area Child Development/Otitis Media Study Group, PA
4642	Paradise	2000	Pittsburgh-area Child Development/Otitis Media Study Group, PA
2819	Wright	1988	Pneumococcal vaccine study in Nashville, TN
1677	Knishkowy	1991	PROD (Promotion of Growth and Development) Program, Western Jerusalem
2579	Teele	1984	The Greater Boston Otitis Media Study Group, MA
4664	Feagans	1994	Three day-care facilities in central Pennsylvania, PA

Table 27 lists the author, year and cohort of 22 prospective cohort studies that were excluded from our analysis because they did not report outcomes for children older than 3 years of age. These 22 studies actually included only 12 cohorts. Relevant findings for seven of these 12 cohorts were included in our assessment.

Findings

Evidence Table 2 presents the study characteristics, population characteristics, risk factors, and outcome findings for the 20 cohort studies that responded to this question. Speech and language outcomes were examined in a total of 12 cohorts of children. Of these 12 cohorts, nine included children primarily from specific populations, such as a particular ethnic or racial group or a particular socioeconomic group. It is also important to reiterate that cohorts were excluded if they consisted exclusively of children with craniofacial defects, primary mucosal disorders, immunodeficiencies, or a genetic disorder. None of the studies specifically assessed children who already had speech, language, or other developmental delays. One of the studies focused specifically on persistent bilateral OM.

Table 28. Definition of Early Life Positive or Negative (more...)

Table 28. Definition of Early Life Positive or Negative OM History in Assessing Long-term Speech and Language Development

ID#	Author Year	Definition of Positive/Negative OM History	Definition of OM	OM Diagnosis Method
877	Black 1993	OM History defined by number of episodes of OM within the first year of life documented by otologic examination. Positive OM History: At least 2 episodes of OM within the first year. A child could receive credit for only one bout of OM within each 29-day period. Negative OM History: Had not experienced otitis media during first year of life	Not provided	Based on otologic examination Examiner not provided
1219	Feagans 1987	OM History described by frequency and duration of OM from 0 to 3 years. They were treated as continuous variables for correlation with outcome. No grouping of children by positive or negative history was done. Frequency was calculated by counting the number of different episodes Duration was calculated by counting the total number of days the child had effusion during the first 36 months of life.	Not provided.	Based on pneumatic otoscopy. Beginning 1978 tympanometry was used to corroborate the diagnosis OM diagnosed by two pediatricians and two nurse practitioners
1255	Fischler 1985	OM History defined by number of OM episodes by age 2 years and over age 2. The study defined three groups of children by OM history by number of episodes before and after age 2. For our assessment, we used groups 1 and 2 as positive OM history, i.e. (>=2 attacks by age 2) and group 3 as negative OM history (<2 attacks by age 2).	Any mention of one of the following: Acute suppurative OM: history of ear pain or fussiness with or without fever or ear drainage (less than 5 days), and physical evidence of redness with or without immobility, bulging, or a small perforation of the TM. Serous OM: history of ear fullness, popping, or hearing loss, or an asymptomatic history; and physical evidence of TM retraction and/or immobility, with or without gray or yellow color or bubbles behind the TM Chronic OM: history of ear drainage and perforation present for more than two weeks; and physical evidence of perforation.	By medical record review of documented physician's clinical diagnosis.
1277	Freeark 1992	OM history severity defined by a) number of separate episodes of OM and b) total number of days of effusion over the first 3 years of life. High OM: above median of OM severity Low OM: below median of OM severity	Not specified	By whom, not specified; How diagnosed, not specified. (OM history obtained from medical records).
1373	Gravel 1992	OM groups were defined by otoscopic histories during the first year of life. OM positive: when bilateral OM was detected at 30% or more of the baby's first year visits. OM negative: when middle ear status was rated as normal in both ears during 80% or more of the first year visits.	Not specified	Pediatric nurse practitioners completed pneumo-otoscopic examinations during each scheduled well-baby visit
4728	Gravel 1996	Same as 1373	Same as 1373	Same as 1373
1435	Harsten 1993	OM groups were defined by the number of AOM episodes during the first 3 years of life. OM positive: developed six or more episodes of AOM during a 12-month period. OM negative: no AOM episodes and less than six other acute respiratory tract infections.	AOM was defined as an acute episode of earache in a child with red bulging eardrum(s) or purulent discharge, occasionally febrile and with signs of upper respiratory tract infection.	By otomicropscopy performed by an otolaryngologist
1623	Kaplan 1973	OM groups were based on age of onset of first episode of otorrhea. The study defined 3 groups: group 1-onset of first otorrhea episode during first year of life; group 2-onset of first otorrhea episode at 2–10 years, and group 3-no history of OM For our assessment, we used group 1 as positive OM history and group 3 as negative OM history.	Used only episodes of OM with otorrhea.	A research nurse visited the cohort children and obtained information concerning middle ear abnormality and upper respiratory tract illness and reviewed medical records for status between visits.
4651	Klein 1988	OM history was measured by time spent with effusion during the first 2 years of life and used a ‘window’ of 23 days to each observation of effusion, whether accompanied by signs of illness or not. It could be shortened or extended by multiple examinations. Group 1: time spent with effusion <32 days during first 2 years of life. Group 2: time spent with effusion between 33–108 days during first 2 years of life. Group 3: time spent with effusion >108 days.	Criteria for effusion: otorrhea, gas-liquid levels visible on otoscopy or marked reduction of mobility. Tympanometric criterion: type B curve.	By pediatricians using pneumatic otoscopy until age 3 and both pneumatic otoscopy and tympanometry in years 4 through 7.
1941	Mody 1999	OM history defined by pneumatic otoscopy findings during first year of life OM positive: children who had 30% or more of the 13 first-year visits with OM bilaterally OM negative: children who had 80% or more of the 13 first-year visits with normal middle ear findings bilaterally.	Used a 9-item otoscopic checklist to determine “clear”, “suspicious,” or “positive” for OM	By trained and validated pediatric nurse practitioner using a pneumatic otoscope under the supervision of a pediatric otolaryngologist. The PNP recorded a description of TM characteristics for each ear, using a 9-item otoscopic checklist and made the determination of “clear,” “suspicious,” or “positive” for OM.
4675	Owen 1996	OME history was measured by days or duration with OME durng the first 3 years of life. Middle ear status was monitored by home visits every 2 to 4 weeks, irrespective of symptoms for the first 3 years of life. At each visit, each ear received a diagnosis of normal or OME. If two consecutive visits showed OME, the intervening days were counted as days with OME. If one visit showed OME and the next normal status, or vice versa, half of the intervening days were counted as days with OME. OME duration was defined as the proportion time a subject spent with OME (total OME days divided by total days) in the period examined. Subjects who experienced 6 continuous weeks of OME in the first year of life were identified as at high risk.	OME diagnosis was based on type B tympanogram or >= 5 acoustic reflectivity or visible purulent otorrhea without an otoscope.	By trained technicians using automated screening tympanometers with a 26 Hz probe tone. Acoustic reflectivity was also measured using acoustic otoscope at 30% of visits.
2135	Paul 1993	OM history was measured by middle ear involvement defined as either the placement of myringotomy tubes or the presence of six or more ear infections treated by a physician before the second birthday by parent report. Positive OM: had middle ear involvement Negative OM: lack of such middle ear involvement	Based on parental reporting	Based on parental reporting
4657	Roberts 1986	OM history was based on total OME duration in days during first 3 years of life. Duration of each episode of unilateral and bilateral was calculated by subtracting the data of onset of OME from the resolution date. Days of total OME was analyzed both as a continuous and categorical variable. Group 1: days with total OME representing the lower third of the subjects Group 2: days with total OME representing the middle third of the subjects Group 3: days with total OME representing the upper third of the subjects.	When middle ear fluid was seen or when the mobility of the tympanic membrane was markedly reduced or absent, OME was diagnosed. Type B tympanograms with a flat or gradually rising shape were considered indicative of OME. Type C tympanograms showing a maximum compliance of less than –100 mm H2O were considered indicative of negative middle ear pressure and of an increase likelihood of effusion	By pediatricians and pediatric nurse practitioners based on pneumatic otoscopy. 60% of the time tympanometry was used to corroborate the OME diagnosis.
4656	Roberts 1989	Same as 4657	Same as 4657	Same as 4657
3118, 4806	Roberts 1988	Same as 4657	Same as 4657	Same as 4657
3117	Roberts 1991	Same as 4657	Same as 4657	Same as 4657
4318	Roberts 1995	Same as 4657	Same as 4657	Same as 4657
2295	Ruben 1997	OM history was based on the findings of pneumatic otoscopy at every scheduled and sick visit during the first year of life The OM – group were those who were bilaterally free of OM at 80% or more of their first year visits and had no more than one episode of OM during the first year. The OM+ group had bilateral OM at 30% or more of their first year visits, and had from 2 to 6 episodes during the first year.	Not specified	Pneumatic otoscopy by a trained and validated pediatric nurse practitioner.
2583	Teele 1990	OM history was measured by the number of days with middle ear effusion (MEE) by age 3 years. Unless documented to be shorter, each episode of MEE lasted 29 days.	Diagnosis of MEE required either visualization of a gas-liquid mixture through an intact tympanic memberane, otorrhea, or marked reduction in mobility of the TM to both positive and negative pressure. Children whose TM(s) showed reduced mobility in response to positive pressure and normal mobility to negative pressure were considered to have only subatmospheric middle ear pressure. The criteria for effusion using tympanometric devices included a tracing that showed no peak or a tracing that sagged below the baseline (for model 1720B)	In private practice, three board-certified pediatricians performed 81% of all exams; in urban health center three board-certified pediatricians performed 66% of all exams. Each center used otoadmittance meters, initially a Grason-Stadler model 1720B. Later, for children 4–7 years, used a Grason-Stadler model 1722. To resolve ambiguous diagnoses, otoadmittance was used sporadically at 0–3 years of age, but frequently at 4–7 years.

For the purpose of responding to this question, we defined early life otitis media as positive otitis media history prior to 3 years of age. Table 28 summarizes the definitions of positive or negative history of otitis media used in these studies and the diagnostic method for OM. The definition of positive or negative OM history varied among the studies. Some studies used percentage of visits during a specified time, some used number of visits with OM, some used days spent with effusion, while others combined various criteria.

The age during which the outcome was measured also varied. This age ranged from one to 3 years. The diagnostic method and the examiner also varied. Several studies based diagnosis on chart review or parents' record. A few studies used pneumatic otoscopy performed by pediatricians, otolaryngologists, or trained professionals.

Table 29. Study Quality for Studies Included in Evidence (more...)

Table 29. Study Quality for Studies Included in Evidence Table on Speech and Language Development

ID	Author(s) Year	Study Quality Score a
877	Black 1993	3 (0,0,0,0,1,1,0,1)
1219	Feagans 1987	5 (1,1,1,0,0,1,0,1)
1255	Fischler 1985	5 (1,1,0,0,1,1,1,0)
1277	Freeark 1992	4 (1,0,0,0,0,1,1,1)
1373	Gravel 1992	5 (1,1,0,0,1,1,1,0)
4728	Gravel 1996	3 (0,1,0,0,0,1,1,0)
1435	Harsten 1993	6 (1,1,1,0,1,1,1,0)
1623	Kaplan 1973	6 (1,1,0,0,1,1,1,1)
4651	Klein 1988	4 (1,0,1,0,0,1,0,1)
1941	Mody 1999	4 (1,1,0,0,0,1,0,1)
4675	Owen 1996	4 (1,1,0,0,0,1,0,1)
2135	Paul 1993	3 (0,0,1,0,0,1,0,1)
4657	Roberts 1986	6 (1,0,1,0,1,1,1,1)
4656	Roberts 1989	6 (1,0,1,0,1,1,1,1)
4806	Roberts 1988	4 (1,0,1,0,0,1,0,1)
3118	Roberts 1988	4 (1,0,1,0,0,1,0,1)
3117	Roberts 1991	2 (0,0,0,0,0,1,0,1)
4319	Roberts 1995	3 (0,0,0,0,0,1,1,1)
2295	Ruben 1997	5 (1,1,0,0,1,1,1,0)
2583	Teele 1990	6 (1,0,1,0,1,1,1,1)

a

The eight components of study quality score are: (1) study cohort clearly defined; (2) subjects assembled at a uniform time point; (3) pathway of subject entry clearly described; (4) complete follow-up achieved; (5) withdrawals/drop-outs described; (6) objective outcomes used; (7) outcome assessment blinded; and (8) extraneous factors adjusted. ‘1’ indicates presence and ‘0’ indicates absence.

The quality of the studies included in the evidence table is summarized in Table 29. Of the 20 studies, five (25 percent) received a score of six of a possible of eight points; four (20 percent) scored five points; six (30 percent) scored four points; four (20 percent) scored three points, and one (5 percent) scored two points.

Table 30. Key Characteristics of the Cohort Studies (more...)

Table 30. Key Characteristics of the Cohort Studies for Question 2

ID#	Author Year	Cohort	Age of OM History	Age at Outcome Measure	Major Outcome statistic	Outcome Measure	Test	Notes
4675	Owen 1996	Texas	0–3 years	5 years	Correlation	Cognition	Stanford Binet
4319	Roberts 1995	North Carolina	2 months- 3 years	12 years	Correlation	Cognition	WISC-R
1623	Kaplan 1973	Alaska	0–1 year	10 years	Mean (range)	Cognition	WISC	Stratified by concurrent hearing status
877	Black 1993	Maryland	8–22 months	4–6 years	Mean (SD)	Cognition	McCarthy
1373	Gravel 1992	New York	0–1 year	4 years	Mean (SD)	Cognition	Stanford-Binet
4657	Roberts 1986	North Carolina	2 months- 3 years	3.5–6 years	Mean (SD)	Cognition	McCarthy
4656	Roberts 1989	North Carolina	2 months- 3 years	8 years	Mean (SD)	Cognition	WISC-R
4651	Klein 1988	Massachusetts	0–2 years	7 years	Multivariate	Cognition	WISC-R
2583	Teele 1990	Massachusetts	0–2 years	7 years	Multivariate	Cognition	WISC-R	Adjusted for SES and gender
3118/4806	Roberts 1988	North Carolina	2 months- 3 years	2.5–8 years	Correlation	Expressive language	Elicited language play
3117	Roberts 1991	North Carolina	2 months- 3 years	4.5–6 years	Correlation	Expressive language	CELF	Stratified by socioeconomic status
4675	Owen 1996	Texas	0–3 years	5 years	Correlation	Expressive language	Goldman-Fristoe test
1255	Fischler 1985	Arizona	0–2 years	6–8 years	Mean (SD)	Expressive language	TOLD
1373	Gravel 1992	New York	0–1 year	4 years	Mean (SD)	Expressive language	SICD-R in months
2135	Paul 1993	Oregon	0–2 years	4 years	Mean (SD)	Expressive language	MLU	Stratified by normal/late talkers
1219	Feagans 1987	Pennsylvania	0–3 years	5–7 years	Multivariate	Expressive language	MLU, Paraphrase	Reported for total group only
2583	Teele 1990	Massachusetts	0–2 years	7 years	Multivariate	Expressive language	WUG test	Mean adjusted for SES and gender
2295	Ruben 1997	New York	0–1 year	2–9 years, yearly	Percent difference	Expressive language	Unknown
1277	Freeark 1992	Michigan	0–3 years	3–4 years	Proportion	Expressive language	Verbal Scale Index	Stratified by Parent Verbal Stimulation (PVS)
4728	Gravel 1996	New York	0–1 year	9 years	Raw data not reported	Expressive language	WRAML	Only statistical testing was reported
3117	Roberts 1991	North Carolina	2 months- 3 years	4.5–6 years	Correlation	Receptive language	CELF	Stratified by socioeconomic status (SES)
3117	Roberts 1991	North Carolina	2 months- 3 years	4.5–6 years	Correlation	Receptive language	PPVT-R	Stratified by SES
4675	Owen 1996	Texas	0–3 years	5 years	Correlation	Receptive language	CAVAT
1255	Fischler 1985	Arizona	0–2 years	6–8 years	Mean (SD)	Receptive language	TOLD
877	Black 1993	Maryland	8–22 months	4–6 years	Mean (SD)	Receptive language	PPVT-R
1373	Gravel 1992	New York	0–1 year	4 years	Mean (SD)	Receptive language	SICD-R in months
2583	Teele 1990	Massachusetts	0–2 years	7 years	Multivariate	Receptive language	WUG test	Adjusted for SES and gender
2295	Ruben 1997	New York	0–1 year	2–9 years, yearly	Percent difference	Receptive language	Unknown
4728	Gravel 1996	New York	0–1 year	9 years	No raw data	Receptive language	CELF-R	Statistical significance only
3118/4806	Roberts 1988	North Carolina	2 months- 3 years	2.5–8 years	Correlation	Expressive speech	Goldman-Fristoe
2135	Paul 1993	Oregon	0–2 years	4 years	Mean (SD)	Expressive speech	Goldman-Fristoe	Stratified by normal/late talkers
2583	Teele 1990	Massachusetts	0–2 years	7 years	Multivariate	Expressive speech	Goldman-Fristoe	Adjusted for SES and gender
2295	Ruben 1997	New York	0–1 year	2–9 years, yearly	Percent difference	Expressive speech	Unknown
3118/4806	Roberts 1988	North Carolina	2 months- 3 years	2.5–8 years	Correlation	Receptive speech	Articulation tests
1941	Mody 1999	New York	0–1 year	9 years	Mean (SD)	Receptive speech	Synthetic speech syllables
1435	Harsten 1993	Sweden	0–3 years	4 years	Proportion	Receptive speech	Linguistic analysis
4728	Gravel 1996	New York	0–1 year	9 years	No raw data	Receptive speech	Pediatric Speech Intelligibility	Only statistical testing result was reported

CAVAT= Carrow Elicited Language Inventory

CELF= Clinical Evaluation of Language Functions

MLU= Mean Length of Utterance

PPVT-R= Peabody Picture Vocabulary Test-Revised

SD= Standard Deviation

SICD-R= Sequenced Inventory of Communication Development-Revised

TOLD= Test of Language Development

WISC-R= Wechsler Intelligence Scale for Children-Revised

WRAML= Wide Range Assessment of Memory and Learning

WUG= “WUG” test (Berko-Gleason)

Table 30 presents a summary of the key characteristics of the 20 cohort studies including the cohort, age of OM history, age at outcome measure, major outcome statistical analysis, outcome measure of interest, diagnostic procedure(s), and notes. The table is organized by the outcome measures for this key question, namely cognition, expressive language, receptive language, expressive speech, and receptive speech. Since multiple outcomes could be included in one study, a study may appear in several rows. The entries in the “Test” column should be interpreted with caution, because subtests versus global tests were not distinguished.

The factor we used to determine which studies to examine further by meta-analysis was the major outcome measure, which indicates the type of statistical measured used. Whenever three or more studies reported the same outcome measure, we considered pooling the data. Based on this strategy, we conducted three possible meta-analyses to derive the pooled difference between positive and negative early-life otitis media for: expressive language development, receptive language development, and development of cognitive verbal intelligence.

Table 31. Meta-Analysis for Expressive Language Development (more...)

Table 31. Meta-Analysis for Expressive Language Development

ID Number	Author-Year	Cohort	Age of OM history	Age of outcome measure	Name of Test	Positive OM History			Negative OM History			Standardized Mean Difference (95% CI)
						N	Mean	SD	N	Mean	SD
1255	Fischler-1985a	Arizona	0–2yr	6–8 yrs	TOLD	33	60.0	20.4	71	64.8	28.8	- 0.18 (-0.59, 0.23)
1373	Gravel-1992	New York	0–1yr	4yrs	SICD-R	8	36.0	5.2	12	39.0	6.2	- 0.49 (-1.40, 0.42)
2135	Paul-1993a	Oregon	0–2yr	4yrs	MLU	8	57.8	3.8	13	54.6	10.7	0.35 (-0.54, 1.24)
Random Effects estimate												-0.14 (-0.49, 0.20)
Test of standardized mean difference equals 0: z=0.82; p=0.413.
Test of heterogeneity: Chi-squared=1.77 (degrees of freedom=2); p-value=0.412.

a

Retrospective-prospective studies.

OM=Otitis media

TOLD=Test of Language Development

SICD-R=Sequenced Inventory of Communication Development-Revised

MLU=Mean Length of Utterance

N=Number of subjects

SD=Standard deviation

CI=Confidence interval

Expressive Language Development Table 31 presents the findings of the three cohort studies that addressed expressive language development. All three studies measured OM history prior to 3 years of age. Although the pooled standardized mean difference showed an increase of about 14.5 percent (95% C.I.: -49.2%, 20.2%) of a standard deviation in the expressive language measure for the group of children with no early-life OM history compared with those with a positive history, this pooled estimate is not significantly different from zero. Thus, the available data do not support the hypothesis that an OM history prior to 3 years of age has an effect on expressive language development. However, the 95% confidence intervals on our pooled results do not exclude a clinically important effect size of almost 0.5, meaning no strong conclusions can be drawn. The Chi-squared test of heterogeneity showed that the standardized difference was not significantly different among the studies. However, the ages at which outcome was measured and the tests used were not uniform across the studies.

Table 32. Meta-Analysis for Receptive Language Development (more...)

Table 32. Meta-Analysis for Receptive Language Development

ID Number	Author-Year	Cohort	Age of OM history	Age of outcome measure	Name of Test	Positive OM History			Negative OM History			Standardized Mean Difference (95% CI)
						N	Mean	SD	N	Mean	SD
877	Black-1993	Maryland	8–22 mos	4–6 yrs	PPVT-R	21	83	17	10	72	18	0.62 (-0.15, 1.39)
1255	Fischler a -1985	Arizona	0–2yr	6–8 yrs	TOLD	33	67	28	71	73	32	-0.19 (-0.61, 0.22)
1373	Gravel-1992	New York	0–1yr	4yr	SICD-R	8	36	5	13	38	5	-0.38 (-1.27, 0.51)
2579	Teele-1990	Boston	0–2yr	3yr	PPVT-R	52	101	17	80	96	15	0.31 (-0.04, 0.67)
Random effects estimate												0.10 (-0.29, 0.49)
Test of standardized mean difference equals 0: z=0.52; p=0.606.
Test of heterogeneity: Chi-squared=6.22 (degrees of freedom=3); p=0.102.

a

Retrospective-prospective study.

PPVT-R=Peabody Picture Vocabulary Test-Revised

TOLD=Test of Language Development

SICD-R=Sequenced Inventory of Communication Development-Revised

N=Number of subjects

SD=Standard deviation

CI=Confidence interval

Receptive Language Development Table 32 presents the findings of the four cohort studies that addressed receptive language development. All four studies measured OM history at less than 3 years of age. Although the pooled standardized mean difference showed an increase of about 10.3 percent (95% C.I.: -28.9%, 49.5%) of a standard deviation of the receptive language measure in the group of children with no early-life otitis media history, this pooled estimate is not significantly different from zero. Thus, the available data do not support the hypothesis that an OM history prior to 3 years of age has an effect on receptive language development. However, the 95% confidence intervals on our pooled results do not exclude a clinically important effect size of almost 0.5, meaning no strong conclusions can be drawn. The Chi-squared test of heterogeneity showed that the standardized mean difference was not significantly different among the studies. However, the age at which outcome was measured and the test used were not uniform across the four studies. In addition, although the racial/ethnic composition was not reported precisely, one of the studies included primarily African-American children, another primarily American Indian children, and the third primarily Caucasian children in private practice.

Table 33. Meta-Analysis for Cognitive Verbal Intelligence (more...)

Table 33. Meta-Analysis for Cognitive Verbal Intelligence

ID Number	Author-Year	Cohort	Age of OM history	Age of outcome measure	Name of Test	Positive OM History			Negative OM History			Standardized Mean Difference (95% CI)
						N	Mean	SD	N	Mean	SD
877	Black-1993	MD	8–22 mos	4–6 yrs	McCarthy	21	46.7	11.5	10	41.0	10.7	0.49 (-0.27, 1.26)
1373	Gravel-1992	NY	0–1yr	4yrs	Stanford-Binet	9	88.3	15.9	13	84.3	9.4	0.31 (-0.55, 1.17)
4657	Roberts-1986	NC	2mos-3yrs	3.5–6yrs	McCarthy	19	52.0	8.0	19	52.0	9.0	0.00 (-0.64, 0.64)
Random effects estimate												0.23 (-0.20, 0.65)
Test of standardized mean difference equals 0: z=1.05; p=0.292.
Test of heterogeneity: Chi-squared=0.99 (degrees of freedom=2); p=0.609.

OM=Otitis media

N=Number of subjects

SD=Standard deviation

CI=Confidence interval

Cognitive Verbal Intelligence Table 33 presents the findings of the three cohort studies that addressed development of cognitive verbal intelligence. All three studies examined cognitive verbal intelligence, and all measured OM history at less than 3 years of age. Although the pooled standardized mean difference showed an increase of about 23 percent (95% C.I.: -20%, 65%) of a standard deviation in the expressive language measure in the group of children with no early life OM history, this pooled estimate is not significantly different from zero. Thus, the available data do not support the hypothesis that an OM history prior to 3 years of age has an effect on the development of cognitive verbal intelligence. However, the 95% confidence intervals on our pooled results do not exclude a clinically important effect size of almost 0.5, meaning no strong conclusions can be drawn. The Chi-squared test of heterogeneity showed that the standardized mean difference was not significantly different among studies. However, the age at which outcome was measured and the test used were not uniform across the studies. Further, two of the study populations were primarily African-American. The third study population was of lower socioeconomic status.