Otitis media can be defined as inflammation of the middle ear mucosa, resulting from an infectious process (Scheidt and Kavanagh, 1986). When the inflammation results in the secretion of effusion, or liquid, into the middle ear cavity, the terms otitis media with effusion (OME) and middle ear effusion (MEE) are often used. Middle ear effusion may be present during the period of acute inflammation (when it is known as acute otitis media), and it may persist for some time after the acute inflammation has subsided (Bluestone and Klein, 1996a).

The prevalence of OME in young children is remarkably high. OME has been described as one of the most common infectious diseases of childhood (Bluestone and Klein, 1996a) and evidence from several large studies supports this conclusion. For example, in a prospective epidemiologic study of 2253 children enrolled by age 2 months, Paradise et al. (1997) reported that nearly 80% of children had at least one episode of OME by 12 months of age; more than 90% had an episode by age 24 months. The mean cumulative percentage of days with MEE was 20.4% between 2 and 12 months of age, and 16.6% between 12 and 24 months. Low socioeconomic status, male sex, and amount of exposure to other children were associated with an increased prevalence of OME during the first 2 years of life (Paradise et al., 1997).

The literature addressing the hypothesis that early recurrent OME poses a threat to children's speech and language development is large and contentious (for reviews, see Stool et al., 1994; Shriberg, Flipsen, et al., 2000). OME has been reported to result in adverse effects, no effects, and small beneficial effects (e.g., Shriberg, Friel-Patti, et al., 2000), sometimes within the same study. Substantive methodological differences may account for much of the disparity in findings, with the method by which OME is diagnosed in different studies being critically important. The gold standard for diagnosing OME is an examination of the tympanic membrane via pneumatic otoscopy, after any necessary removal of cerumen, to determine whether indicators of effusion such as bulging, retraction, bubbling, or abnormal mobility are present (Stool et al., 1994; Bluestone and Klein, 1996b). Behavioral symptoms such as irritability are neither sensitive nor specific to the condition, and the validity of parental judgments concerning the frequency or duration of episodes of OME is poor even when repeated feedback is provided (Anteunis et al., 1999). In addition, it is important that OME be documented prospectively rather than via retrospective reports or chart reviews, because a substantial percentage of apparently healthy and symptom-free children are found to have OME on otoscopic assessment (Bluestone and Klein, 1996a).

A second important difference among studies of OME and speech development is the extent to which hearing levels are documented. The hypothesis that OME poses a threat to speech or language development is typically linked to the assumption that effusion causes conductive hearing loss, which prevents children from perceiving and processing speech input in the usual fashion (e.g., K. Roberts, 1997). However, the presence of effusion is a poor predictor of hearing loss. Although hearing thresholds for the majority of children with MEE fall between 21 and 30 dB (mild to moderate degrees of impairment), thresholds from 0 to 50 dB are not uncommon (Bess, 1986). Hearing thresholds must be measured directly to determine whether OME has effects on development independent of its variable effects on hearing (e.g., Shriberg, Friel-Patti, et al., 2000).

Studies also vary with respect to their ability to separate the contribution of OME to poor developmental outcome from the effects of other variables with which OME is known to be associated, such as sex and socioeconomic status. As noted earlier, OME is significantly more prevalent in males than in females, and in children from less privileged backgrounds than in their more privileged counterparts (Paradise et al., 1997; Peters et al., 1997). Statistical procedures are necessary to control for such confounding in order to distinguish the effects of OME from those of other variables. Several recent studies have shown that after controlling for socioeconomic confounds, OME accounts for little if any of the variance in developmental outcome measures (e.g., J. E. Roberts et al., 1998; Paradise et al., 2000).

Finally, studies have also differed substantially in the measures used to document the outcome variable of speech development and in the extent to which effect sizes for significant differences on outcome measures are reported (cf. Casby, 2001). No accepted standard metric for speech delay or disorder currently exists, although the Speech Disorders Classification System developed by Shriberg et al. (1997) represents an important advance toward meeting this need. Instead, the effects of OME on speech have been sought on a wide range of articulatory and phonological measures, not all of which are known to be predictive of eventual speech outcome (e.g., Rvachew et al., 1999).

When these cautions are borne in mind, the literature on early recurrent otitis media and speech development suggests converging evidence that OME in and of itself has a negligible relationship to early speech development. Several prospective investigations have shown little or no relationship between cumulative duration of otitis media (documented otoscopically) and measures of speech production in otherwise healthy children. In a longitudinal study of 55 low-SES children, J. E. Roberts et al. (1988) found no correlation between OME and number of consonant errors or phonological processes on a single-word test of articulation at ages 3, 4, 5, 6, 7, or 8 years. Shriberg, Friel-Patti, et al. (2000) examined ten speech measures derived from spontaneous speech samples obtained from 70 otherwise healthy, middle to upper middle class 3-year-olds who were classified according to the number of episodes of OME from 6 to 18 months of age; only one significant speech difference was found, in which the group with more OME paradoxically obtained higher intelligibility scores than the group with fewer bouts of OME. Paradise et al. (2000) likewise found no relationship between cumulative duration of MEE and scores on the Percentage of Consonants Correct–Revised measure (PCC-R; Shriberg, 1993) in 241 sociodemographically diverse children at age 3 years. Paradise et al. (2001) reported that PCC-R scores from children with even more persistent MEE from 2 to 36 months of age did not differ significantly from those of children with the less persistent levels of effusion reported by Paradise et al. (2000). Further, children with persistent and substantial MEE who were randomly assigned to undergo prompt tympanostomy tube placement had no better PCC-R scores at age 3 than children who underwent tube placement after a delay of 6–9 months, during which their MEE persisted (Paradise et al., 2001). These findings of little or no relationship between OME and speech development mirror those of several recent reports showing negligible associations between early OME and later oral and written language performance (Peters et al., 1997; Casby, 2001).

By contrast with these negative findings concerning the impact of OME, several studies in which hearing was documented showed poorer speech outcomes for children with elevated hearing thresholds. In a sample of 70 middle to upper middle class 3-year-olds who received otoscopic evaluations every 6 weeks and hearing evaluations every 6 months between 6 and 18 months of age, Shriberg, Friel-Patti, et al. (2000) reported that children with hearing loss, defined as average thresholds (20 dB (HL) during one evaluation between 6 and 18 months of age, had a significantly increased risk of scoring more than 1.3 standard deviations below the sample mean on several percentage-consonants-correct metrics. Shriberg et al. note the need for some caution in interpreting these findings, given that increased risk was not found across all speech metrics and that confidence intervals for risk estimates were wide. In addition, the results of structural equation modeling suggested that hearing loss did not operate directly to lower speech performance, but rather was mediated significantly by language performance, providing another indication of the need for multifactorial approaches to identifying the factors and pathways involved in normal and abnormal speech development.

Although the best available current evidence suggests that OME itself does not represent a significant risk to speech development in otherwise healthy children, the question of whether OME may contribute independently to outcome when it occurs in conjunction with other risk factors or health conditions (e.g., Wallace et al., 1996; J. E. Roberts et al., 1998; Shriberg, Flipsen, et al., 2000) remains open. Additional investigations that include prospective otoscopic diagnosis of OME; frequent and independent assessments of hearing; valid, reliable assessments of both medical and sociodemographic risk factors; and a multifactorial analytic strategy will be needed to answer this question.

See also otitis media: effects on children's language.