Objective. To examine whether otitis media with effusion (OME) and associated hearing loss during the first 4 years of life are related to the language development and academic achievement of children between 4 years of age and second grade.
Methods. In a prospective cohort study, 83 black children, primarily from low-income families and recruited from community-based child care programs, were examined repeatedly between the ages of 6 months and 4 years for the presence of OME and hearing loss, both when well and when ill with OME. Children’s child-rearing environments at home were assessed annually from infancy through second grade, whereas children’s language and academic skills were assessed repeatedly between 4 years of age and second grade.
Results. We did not find in our longitudinal analyses a relationship between OME and hearing loss during the first 4 years of life and later academic skills in early reading and recognition of words heard. We did find that children with greater incidence of OME and hearing loss during the first 4 years of life scored lower in verbal math problems between kindergarten and second grade, even after partialing out important background factors. Children with more OME tended to score lower in math at the younger ages but caught up once they entered school. Follow-up analyses indicated also that children with more OME during the first 2 years of life scored lower in expressive language during the preschool and early elementary school years but caught up by second grade. In contrast, children from homes that were rated as more stimulating and responsive scored higher on every measure of language and academic skills than did children from less responsive homes. The home environment was related more strongly to all of the outcomes examined than was OME or hearing loss.
Conclusions. There was no evidence of a significant relationship between a history of OME or hearing loss and children’s later academic skills in reading or word recognition during the early elementary school years. Children with greater incidence of OME and hearing loss scored lower in math and expressive language at the younger ages but caught up in math with their peers on entering school and in expressive language by second grade. Furthermore, a child’s home environment was more related to early math and expressive language skills than was OME or hearing loss, and the home environment continued to be predictive of all of the language and academic outcomes through second grade. These study results should be interpreted cautiously when generalizing to other populations.
Although several recent studies have failed to relate an early childhood history of recurrent or persistent otitis media with effusion (OME) to language development during the first 3 years of life,1–4 the linkage of a history of early OME to later language and academic skills continues to be controversial.5–7 One important reason to examine this linkage is that otitis media, one of the most frequent illnesses of early childhood,8,9 occurs during the years that are critical for language learning and that provide the basis for later literacy and math skills. As a child listens to the stream of speech in the environment and attaches meaning to these sounds, he or she learns to abstract the rules for language. A child with an episode of OME often experiences a mild to moderate fluctuating hearing loss, thus receiving partial or inconsistent auditory cues, which may make speech more difficult to detect and/or to filter from background noise. It has been hypothesized that the resulting misperception or missing of words may affect the input to the knowledge base or to the neural substrate on which language learning is built. It also has been proposed that any difficulties attributable to OME-associated hearing loss may become particularly evident when a child reaches school age and faces the challenges of the school environment. Academic skills, particularly in reading and other language-based subjects, may be affected when there is a high demand for attention to verbally presented information.
A second justification for continuing to study the impact of OME on later language and academic skills has to do with the inconsistent study findings even among the prospective cohort studies. Some researchers have found that a history of OME in early childhood results in lower scores on measures of language skills during the early elementary school years10–13 and in lower scores on later academic achievement measures in school.14–17 Other studies have not found these associations for language12,17–21 or for academic achievement22–24 during the school-age years.
Third, there have been limitations in study design even among the prospective cohort studies of the OME developmental linkages. Few studies have examined hearing loss, which can vary greatly with each OME episode, as a predictor variable, although the linkage between OME and developmental outcomes is believed to result from the degree of hearing loss. The hearing loss associated with an OME episode can range from no loss to as much as a 50-decibel (dB) hearing level (HL; a moderate loss). Previous studies also have not considered the role of the caregiving environment, a well-documented factor in children’s language development and later academic skills.25–27 Finally, few studies have examined the same outcomes longitudinally over time. The finding of a significant relationship at one point in time may have little relevance if it is no longer apparent when the child is older. It is only with longitudinal analyses of the same variables that both the level and the growth in language development, as related to OME history, can be studied.
We reported in our previous prospective studies28,29 that we did not find a direct association of OME or related hearing loss to language and cognitive development at 1 or 2 years of age. However, children with more frequent OME and associated hearing loss tended to have less responsive home and child care environments, factors that were linked to lower language skills at 1 and 2 years of age. In a follow-up study,30 we did not find a significant relationship between children’s early OME history or hearing loss and their language skills during the preschool years (ages 3–5). However, we did find that children with more frequent OME and associated hearing loss scored lower in school readiness skills in math and in recognizing incomplete words at entry into kindergarten. A child’s home environment was more strongly related to these school readiness skills than was OME or hearing loss. The current study follows the same children to determine whether children’s OME history and hearing sensitivity during the first 4 years of life are related to the level of and growth in their language development and academic achievement from 4 years of age through second grade. We also examined the relationship of a history of OME and hearing loss to the responsiveness and supportiveness of the child’s home environment using a standard measure that has been shown to be related to the development of children’s language and academic skills.25–27 Therefore, inclusion of a standard measure of the home environment is helpful in interpreting the size of any effects that might relate to a history of OME or associated hearing loss.
Eighty-three black children (39 boys, 44 girls) who were in a longitudinal study of children’s otitis media and development were participants in this study. Fifty-seven of the study children participated in an earlier study of OME experience and 1-year outcomes,28 82 of the children were in a study of OME and 2-year developmental outcomes,29 and all of these children participated in an earlier study of OME and 3- to 5-year developmental outcomes.30 Children were enrolled into this study without previous knowledge of their ear history. They were recruited from 9 center-based child care programs at ages 6 to 12 months (mean age: 8.2 months) during a 20-month period. No child had any known medical or genetic abnormalities when entering the study. Sixty-five children were born at full gestational age (>37 weeks); 11 were born at 37 weeks, 6 at 36 weeks, and 1 at 30 weeks.
When children entered the study, 74.7% of families were classified as low-income based on whether family income was <185% of the federal poverty threshold (<$20 609 for a family of 3). Sixty-eight percent of the primary guardians were single. Seven of the children were not living with their biologic mother at 1 year of age: 3 lived with a grandmother, 3 lived with a foster or adoptive mother, and 1 lived with a father. The highest level of education of the child’s caregivers was less than a high school degree (30.1%), high school degree (26.5%), some college or other training after high school (32.4%), or a college degree (11%). At the time of the children’s birth, the mothers’ mean age was 24.2 years (standard deviation [SD] = 5.4) and ranged from 14 to 38 years. The Academic Affairs Institutional Review Board at the University of North Carolina at Chapel Hill approved study protocols each year. The child’s parent or guardian provided informed consent.
Otitis Media Documentation
Children’s ears were examined between study entry and 4 years of age by 2 pediatric nurse practitioners who were trained by a pediatrician in pneumatic otoscopy. Weekly ear examinations were provided during the first 15 months of the study and then, as a result of a protocol change, biweekly for the remainder of the study. Children who left their original child care sites remained in the study and had their ears examined monthly. Children were seen for an average of 71.7 times between study entry and 4 years of age (SD = 18.4). The interval between ear observations was approximately 1 week in length for 13.3% of the observations, 2 weeks for 54.8%, monthly for 28.9%, and 2 months for 2.9%.
Diagnosis of OME was based on pneumatic otoscopy; tympanometry (226 Hz probe tone) corroborated the OME diagnosis. OME was diagnosed when fluid was observed in the middle ear and the tympanic membrane was immobile. A flat (type B) tympanogram was characterized by low static admittance (≤0.2 milliohm) or by no discernible pressure peak. When the findings of otoscopy and tympanometry did not agree, the otoscopic diagnosis was used. On the basis of 12 285 ear examinations, excluding examinations when tympanostomy tubes were present, there was a 91% agreement between pneumatic otoscopy and tympanometry for judgments of membrane mobility. Interobserver agreement levels between the primary nurse practitioner and a board-certified otolaryngologist was very high (κ = 0.87), as has been previously reported.28–30 Parents were informed of findings after each ear examination, and children’s primary health care providers were informed of all treatments prescribed for study children. Children’s primary care providers made decisions regarding referrals for additional evaluation and consideration of tympanostomy tube placement. Additional information about the OME diagnostic and treatment procedures can be found in previous reports.28–30
The percentage of time from study entry until 4 years of age during which each child was observed to have unilateral, bilateral, total (unilateral or bilateral), or no OME was computed. For each OME episode, the date of OME onset was subtracted from the date of resolution of OME. The date of OME onset was calculated as midway between the last day on which the ears were observed as normal and the first day OME was present. OME resolution was computed as midway between the last day on which OME was observed and the first day ears were normal.
Hearing sensitivity was assessed by using visual reinforcement audiometry (VRA) between 6 months and 2.5 years and using play audiometry (PLAY) between 2.5 and 4 years. For VRA, randomized frequency-modulated pure tones at 500, 2000, and 4000 Hz were presented through a loudspeaker in a calibrated sound field. For PLAY, pure tones at 500, 1000, 2000, and 4000 Hz were presented via earphones. Children’s hearing was tested 1) at entry into the study and every 3 months thereafter; 2) during weeks 1, 4, 7, and 13 after diagnosis of OME; and 3) after a change in ear status (eg, bilateral to unilateral OME, bilateral OME to normal). Children’s hearing was tested an average of 22.1 times (SD = 4.7) between study entry and 4 years of age. Hearing testing was done in a mobile testing van that housed a single-wall sound suite. Sound level measurements were conducted periodically at each child care site, and mean ambient noise levels were consistently within allowable levels.30 The audiologist and assistant were blind to the child’s ear status. Additional details of the hearing testing procedures performed are described in previous studies.28–30
A summary index of hearing loss was calculated by averaging thresholds in each session. For VRA, a mean of thresholds ≥25 dB HL was considered indicative of significant hearing loss. For PLAY, the thresholds for left and right ear were averaged. Significant hearing loss was indicated when the mean of these thresholds at 500, 1000, 2000, and 4000 Hz was ≥20 dB HL. The computation for hearing loss when VRA was administered is similar to our previous computational methods.28 The percentage of days during which each child had significant hearing loss was then calculated, using the method described above for calculating length of OME.
Language and School Readiness Measures
Children’s language and early literacy and math skills were assessed repeatedly once a year from 4 years of age through second grade. Children were tested at 4 years of age and at kindergarten entry with the Clinical Evaluation of Language Fundamentals-Preschool (CELF-P)31 and at the end of kindergarten and in second grade using the Clinical Evaluation of Language Fundamentals (CELF-3),32 instruments that measure overall receptive and expressive language. Overall receptive and expressive language scores were computed. Both the CELF-P and CELF-3 have adequate levels of reliability and validity.
The children also were repeatedly given 3 subtests from the Woodcock-Johnson Psychoeducational Battery (WJ).33 Letter-Word Identification, which assesses reading identification skills in identifying isolated letters and words (eg, “What is the name of this letter?” while pointing to the letter “O”), was administered at kindergarten entry, at the end of kindergarten, and at the end of first and second grades. Applied Problems, a subtest that assesses skills in analyzing and solving verbal math problems (eg, showing a picture of 4 unhatched and 2 hatching eggs and asking, “How many eggs have not hatched?”), was given at kindergarten entry and at the end of first and second grades. The Incomplete Words subtest, which measures a child’s ability to identify a complete word after hearing the word on audiotape with one or more phonemes missing (eg, “otato” for “potato”), was administered at kindergarten entry and at the end of kindergarten and second grade. Standard scores, called W scores, for Letter-Word Identification, Applied Problems, and Incomplete Words were computed. W scores are based on the Rasch ability scale and are centered on a value of 500, which is the approximate average performance of a beginning fifth grader. During the 7 years of the study, the annual language and academic assessments were administered by 1 of 7 trained examiners with expertise in speech and language assessment. All of the examiners remained blind to the children’s OME and hearing history. Tests were given in the sound room of a mobile test van, in a sound room at a university research center, or at the child’s school.
The overall quality and responsiveness of the home environment was assessed at 9, 18, 30, and 42 months of age using The Home Observation for Measurement of the Environment-Inventory for Infants or Preschoolers (HOME).34 Subscales of the HOME measure the primary caregiver’s emotional and verbal responsivity, acceptance of the child’s behavior, organization of the environment, academic and language stimulation, and maternal involvement with the child. The HOME has high levels of validity and reliability.35 The HOME total score used in the analyses was computed by adding together the item scores at each age (9, 18, 30, and 42 months) and averaging the 4 scores. During the 5-year period, home visits were conducted by 2 trained nurse practitioners and 2 speech-language pathologists who had achieved interrater agreement of at least 95% before collecting the data.
In this study, both descriptive and inferential analyses tested the associations between OME and/or hearing loss, quality of the home environment, and children’s language and academic outcomes. Because both OME and hearing loss were skewed within this sample, a rank transformation was applied to all of the OME and hearing loss variables. These rank-transformed OME and hearing loss variables were used in all analyses. One child did not have complete hearing data, and this child was included only in the OME analyses. In the descriptive analyses, the linear association among the measures of otitis media, hearing loss, family characteristics, and child outcomes was examined using 0-order correlations. Pearson correlations were run between rank of total OME, rank of hearing loss, ratings of the responsiveness/supportiveness of the family environment, and the child outcomes both at each age and averaged across the period from 9 to 42 months. Longitudinal analyses tested whether OME or associated hearing loss during the first 4 years of life was related to developmental patterns of children’s academic and language skills between 4 years of age and second grade. Hierarchical linear models were fit to each of the 5 outcomes: CELF Receptive Language, CELF Expressive Language, WJ Incomplete Words, WJ Letter-Word Identification, and WJ Applied Problems. The models estimated individual linear growth curves, estimating random effects intercepts and slopes for each child for each outcome. The quadratic group growth curves described patterns of change over time as a function of family variables, the child’s gender (male = 1, female = 0), and either ranked proportion of time with OME (entry to 4 years of age) or time with hearing loss (study entry to 4 years of age). The family variable was composed of 2 variables: mother’s (or primary caregiver’s) education (highest year attained) and the child’s averaged HOME scores across the infancy and preschool-age periods (proportion of items passed). Models tested whether each predictor predicted scores at the average age (main effects) or patterns of linear change (age interactions) or nonlinear (age2 interactions) change.
Post hoc follow-up analyses were conducted to determine whether our decision to examine OME and associated hearing loss during the first 4 years might have resulted in ignoring important associations between OME in infancy and later learning. The first 2 years of life is a time period often viewed as critical for language learning. To address this concern, we ran 2 sets of analyses. First, to determine whether either OME or hearing loss was significantly associated with an outcome, we asked whether the amount of OME or hearing loss during the first 2 years or between 2 and 4 years accounted for this finding. The longitudinal analyses were rerun, replacing the variable representing OME or hearing loss between 6 and 48 months with 2 variables representing OME or hearing loss between 6 and 24 months and between 24 and 48 months. Second, we conducted all analyses a second time using OME or hearing loss between 6 and 24 months instead of OME or hearing loss between 6 and 48 months as the predictor of interest. All covariates were included in both sets of analyses.
Otitis Media Experience and Hearing Sensitivity
Children experienced total OME (combined unilateral and bilateral) 63.7% of the time (SD = 26.6) between enrollment and 2 years, 17.9% of the time between 2 and 4 years of age, and 35.6% of the time (SD = 21.8) between enrollment and 4 years of age. Children experienced hearing loss 31.5% of the time (SD = 29.1) between enrollment and 2 years, 12.4% of the time (SD = 19.6) between 24 and 48 months of age, and 19.8% (SD = 20.8) of time across the entire study period.
The amount of OME and hearing loss was highly correlated within and across time. The proportion of days with OME and with hearing loss was highly correlated during the first 2 years (r = 0.71; P < .001), between 2 and 4 years of age (r= 0.68; P < .001), and across the entire study period (r = 0.75; P < .001). Children who experienced more total OME between ages 6 months and 2 years had more OME between 2 and 4 years (r = 0.61; P < .001) and between entry and 4 years (r= 0.92; P < .001). Similarly, children who had more hearing loss before their second birthday had more hearing loss between their second and fourth birthdays (r = 0.52; P < .001) and across the entire study period (r = 0.90; P < .001).
Our previous paper28 reported the breakdown by year for children’s OME and hearing experience from 1 to 4 years of age. The means for the family variables (maternal education and HOME) and the child’s gender are also shown in Table 1.
Otitis Media, Hearing, Quality of Child-Rearing, and Developmental Outcomes
As shown in Table 2, the outcome measures were at the population mean for expressive language on the CELF Expressive Language from 4 years of age through first grade and were somewhat below the population mean at second grade. The CELF Receptive Language was somewhat below the population mean for all ages studied. The WJ standardized scores for both Letter-Word Identification and Applied Problems were below the population mean at kindergarten entry and at the mean for the other grades tested, whereas the WJ standardized scores for Incomplete Words were below the mean at all ages.
The outcomes showed moderate to high correlations over time, which is anticipated given repeated assessments of the same or related constructs (Table 3). OME and hearing loss showed only mild associations with some of the child and family covariates (Table 3). Slightly more OME and hearing loss were experienced by children living in less responsive/stimulating home environments.
General linear mixed models (called hierarchical linear models by psychologists) were fit to the repeated assessments of each outcome. These models estimated individual growth curves by estimating a separate intercept and linear slope as random effects for each child. The parameter coefficients and their standard errors (SEs) are shown in the first 3 rows of Table 4. The group growth curves were estimated from the individual growth curves and described level and patterns of change over time in the outcome as a function of age, family variables, the child’s gender, and either ranked proportion of time with OME or with hearing loss 6 months to 4 years of age. The models included a quadratic slope for age to test whether rates of change over time were nonlinear. Two family variables, maternal education and the quality of the family environment (HOME), as well as child’s gender and either OME or hearing loss 6 months to 4 years of age, were included as main effects and crossed with both age and age2. All continuous predictors were centered at the sample mean to enhance interpretation of parameter coefficients. A block test tested the extent to which a set of predictors added to the model. We tested the contribution of the main effects and interactions with age and age2 for 3 blocks: family variables (maternal education and HOME), gender, and either OME or hearing loss 6 months to 4 years. These regression coefficients and block tests are listed in Table 4. All regression coefficients from the analyses including OME 6 months to 4 years are shown in Table 4, whereas coefficients and block tests for the analysis including the hearing loss block 6 months to 4 years are also shown. This is because the other parameters varied very little between the 2 analyses. Partial correlations were also computed between each of the outcome measures at each age and OME 6 months to 4 years (partialing out the child’s gender, HOME, and maternal education), hearing loss 6 months to 4 years (partialing out the child’s gender, HOME, and maternal education), and HOME (partialing out the child’s gender, maternal education, and OME) to provide a more interpretable statistic to describe the associations than the regression coefficients (Table 5). The first outcome to be analyzed was the development of reading skills as measured by the WJ Letter-Word Identification subtest. The individual and group growth curve parameters are listed in the first column of Table 4. The first 4 rows describe individual patterns of growth in reading. Children showed significant linear (β = 34.0; P < .001) and quadratic growth (β = −2.49; P < .001) between the prekindergarten assessment and the second grade assessment, with significant individual differences in both the intercept (SD = 14.6, χ2[1, n = 75] = 25.30, P < .0001) and linear slope (SD = 4.4, χ2[1, n = 75] = 4.67, P = .02). The quadratic curvature suggests that children as a whole acquired reading skills more quickly in kindergarten and first grade than in second grade. The intercept and slope were not significantly correlated, suggesting that the rate at which children acquired reading skills was not related to their initial levels.
The remaining rows in Table 4 describe factors associated with individual differences in the development of these reading skills. Listed in the table are block tests for the 2 family variables, maternal education and HOME, gender, and either OME or hearing loss. Within each block, we simultaneously tested the main effects and interactions with age and age2. Individual parameter estimates were interpreted only when their block was significant. Only the family block significantly contributed to predicting development of reading skills (F[6,138] = 5.50; P < .0001), with higher reading scores observed across time for children from families rated as providing more responsive and stimulating home environments (β = 0.95; F[1,138] = 15.10; P < .001). Neither gender nor OME was significantly associated with reading in this model. Hearing loss also did not contribute significantly when examined in a separate model.
The outcomes for Applied Problems, Incomplete Words, CELF Receptive Language, and CELF Expressive Language are shown in the next 4 columns of Table 4. The children showed substantial individual variability in the development of all outcomes. The family block was associated significantly with all outcomes: WJ Applied Problems (F[6,67] = 4.90; P < .001)], WJ Incomplete Words (F[6,54] = 3.59; P = .004), CELF Receptive Language (F[6,129] = 3.59; P = .003), and CELF Expressive Language (F[6,129] = 5.50; P < .0001). The HOME positively related to the development of all 4 developmental outcomes and related to the faster acquisition of expressive language after children entered school. The row labeled “HOME” lists the main effect coefficient for the HOME in each analysis, and the 2 labeled “HOME × age2” lists the coefficients for this interaction. As shown in Table 4, all HOME main effects and 1 HOME × age2 interaction are significantly different from 0. Gender related to children’s language skills, CELF Receptive Language (F[3,129] = 3.34; P = .02), and CELF Expressive Language (F[3,129] = 2.88; P < .04). Boys tended to score lower on both language scales (receptive: β = −7.8; expressive: β = −5.6) but showed faster acquisition of receptive language after entering school (β = 1.01).
OME and hearing loss were related to developmental patterns for 1 outcome. The development of math skills was related to both OME 6 months to 4 years (F[3,67] = 4.32; P < .01) and hearing loss 6 months to 4 years (F[3,67] = 4.41; P < .001). Amount of OME was significantly related to linear (β = 0.04) and quadratic change (β = −0.03) in math skills and over time was associated negatively with this measure of math skills. Similarly, amount of hearing loss in the first 4 years was significantly related to quadratic change (β = −0.03). Overall, children with more OME and hearing loss from 6 months to 4 years of age tended to score nonsignificantly lower on math at the younger ages but caught up once they entered school. This finding in math is shown in Fig 1 by graphing the predicted growth curves for children whose hearing loss experiences was 1 SD above and below the mean for WJ Applied Problems. Fig 1 shows the anticipated growth in the development of math (WJ Applied Problems) for children from prekindergarten through second grade (approximate ages 5–8 years) whose hearing loss 6 months to 4 years scores were 1 SD above the sample mean (hearing loss present 42% of the time between entry to project and 48 months of age) and 1 SD below the sample mean (hearing loss present 2% of the time between study entry and 48 months of age). The discrepancy between the 2 curves is never large and becomes smaller over time.
In contrast, the family block was related significantly to all outcomes. Children from homes that were rated as more stimulating and responsive showed better academic skills and language overall (see main effect for HOME in Table 4). For example, Fig 2 shows the predicted growth curves in math (WJ Applied Problems) for children from kindergarten entry through second grade whose HOME scores are 1 SD above (passing 88% of HOME items across time) or below (passing 68% of HOME items across time) the mean. Comparisons of Fig 1 versus Fig 2 demonstrate the relative magnitude of the OME or hearing loss as compared with HOME effects.
Two sets of post hoc analyses were conducted. First, we asked whether OME and hearing loss during the infant or preschool years accounted for the observed association with the acquisition of math skills. The longitudinal analyses were run again, replacing OME 6 to 48 months with OME 6 to 24 months and OME 24 to 48 months and replacing hearing loss 6 to 48 months with hearing loss 6 to 24 months and hearing loss 24 to 48 months. Neither infant nor preschool OME or hearing loss significantly contributed to predicting development of math when considered jointly.
Second, we asked whether failure to observe substantial associations between OME and associated hearing loss and later learning was attributable to our decision to measure OME and hearing loss through the child’s fourth birthday instead of only during the first 2 years of life. Follow-up analyses were conducted, using the same model as is reported in Table 4 except that we substituted OME and hearing loss for 6 to 24 months for OME and hearing loss for 6 to 48 months. In general, the findings were very similar and regression coefficients changed only slightly. The acquisition of math skills was significantly associated with both OME 6 to 24 months (F[3,68] = 3.23; P = .03) and hearing loss 6 to 24 months (F[3,68] = 2.74; P = .049). Children showed significantly more initial gains and leveling off in their math skills at the entry to school when they had experienced more OME (β = −0.03; SE = 1; P = .04). The same trend was observed with hearing loss, although the hearing loss × age2 parameter estimates were not significantly different from 0 (β = −0.02; SE = 1; P = .09). However, the OME block was now significant in analysis of expressive language (F[3,132] = 3.15; P = .027), whereas this block has been “marginal” in previous analyses of OME 6 to 48 months (F[3,129] = 2.60; P = .055). Overall, children with more OME from 6 months to 2 years of age tended to score lower on expressive language (β = −0.09; SE = 0.04; P = .03) at the younger ages but were catching up by second grade. They showed more linear gains (β = 0.05; SE = 0.02; P = .02) and quadratic change (β = 0.02; SE = 0.01; P = .006). Figure 3 shows the comparable growth curves for children from prekindergarten through second grade (approximately 5–8 years) for children whose OME 6 months to 2 years was 1 SD above the mean (OME present 57% of the time) or below the mean (OME present 13% of the time) on the CELF Expressive Language. Figure 4 shows the predicted growth curves on the CELF Expressive Language for children from prekindergarten through second grade whose HOME scores are 1 SD above (passing 88% of HOME items across time) or below (passing 68% of HOME items across time) the mean. As with the math growth curves, we see that this family characteristic accounts for more variability in the acquisition of math skills than does OME by comparing Figs 3 and 4.
The 0-order correlations and partial correlations between each of the outcomes at each test point and children’s OME 6 months to 4 years, hearing loss 6 months to 4 years, and HOME are shown in Table 5. Children with more total OME 6 months to 4 years scored lower on WJ Applied Problems at kindergarten entry (r = −0.29) and on the CELF Expressive Language at the end of kindergarten (r = −0.25), even after partialing out the child’s gender and family background factors. Similarly, for hearing loss 6 months to 4 years, children with more hearing loss scored lower on WJ Applied Problems at kindergarten entry (r = −0.33) and on WJ Incomplete Words (r = −0.27) at kindergarten entry, even after considering the child’s gender and family background factors. In contrast for the HOME, children with lower HOME scores scored lower on 15 of the 18 language and academic outcomes examined (r = 0.24–0.43; mean: 0.34). The partial correlations between HOME and the same outcome variables that were significant for OME or hearing loss were 0.33 for WJ Applied Problems at kindergarten entry, 0.43 for WJ Incomplete Words, and 0.24 for CELF Expressive Language, even after partialing out the child’s gender, other family background factors, and OME (Table 5 for 0-order and partial correlations).
Next, we examined the 0-order correlations and partial correlations between each of the outcomes at each test point and children’s OME 6 months to 2 years and hearing loss 6 months to 2 years. Children with more total OME 6 months to 2 years scored lower on WJ Incomplete Words at kindergarten entry (r = −0.25) and on the CELF Expressive Language (r = −0.26) at the end of kindergarten, even after partialing out the child’s gender and family background factors. Similarly, for hearing loss 6 months to 2 years, children with more hearing loss scored lower on WJ Applied Problems at kindergarten entry (r = −0.25), even after considering the child’s gender and family background factors.
Longitudinal analyses of children’s language and early academic skills between 4 years of age and second grade indicated that there was not a significant relationship between a history of OME or hearing loss during the first 4 years of life and children’s later academic skills in reading or word recognition. Although children who experienced more OME and hearing loss in early childhood scored lower in expressive language and verbal math problems at the younger ages, they tended to catch up in math with their peers on entering school and in expressive language by second grade. In contrast, the responsiveness and support of children’s home environments was related more strongly to the expressive language and math scores than was OME or hearing loss and was predictive of almost all of the language and academic outcomes from 4 years of age through second grade.
We found in this study a significant relationship between a history of early OME from 6 months to 2 years of age and lower expressive language scores at the end of kindergarten, but the relationship was no longer significant by second grade. Thus, children caught up in their expressive language skills by second grade. Possibly, expressive language skills may be vulnerable to a history of OME (or to some other factor related to both OME history and expressive language skills); a history of OME does not contribute to the development of expressive language by the time children are in second grade. Several cautions about this finding should be considered. First, although we found an association between an OME history and children’s expressive language at the end of kindergarten, these correlations were not significant at the other 3 ages tested (4 years, before kindergarten entry, and second grade), ranging from −0.02 to −0.14 after partialing out the background variables. Thus, longitudinal analyses provide important information in understanding whether both the level and the growth in language development relate to a previous OME history. Second, we did not find evidence of this linkage with hearing loss only for OME, although hearing loss rather than OME is the variable that we hypothesized to affect later language development. Third, we found that OME from 6 months to 2 years of age, and not 6 months to 4 years of age, was related to later expressive language development; the results had been “marginal” in previous analyses using OME 6 months to 4 years as a predictor. The significant association for the earlier time period may suggest that children are more susceptible to any developmental differences in expressive language when the OME occurs during the first 2 years of life. Furthermore, these findings from longitudinal analyses of language skills assessed at ages 4 through 8 years are not consistent with our previous studies in which we found no association between OME and children’s language skills during infancy or the preschool years.28–30
These findings are consistent with a small number of earlier prospective studies that did find a significant relationship between a history of OME and children’s later language skills. In 1 recent randomized study, Maw et al10 reported higher receptive and expressive language for children (mean age: 3.5 years) who had prompt rather than delayed insertion of tympanostomy tubes. Within 18 months after the treatments, however, there were no longer differences in language skills between the groups. Other studies have reported a relationship between OME history and specific aspects of vocabulary in 7-year-olds,11 grammar in 7-year-olds,12 and language use in 5- and 7-year-olds.13 The present study findings are somewhat difficult to compare to previous studies that examined direct OME language sequelae, as most of the previous studies did not consider the important roles of hearing status or other child and environmental factors that may affect children’s language development.
These findings are inconsistent with a number of studies12,17–19 that did not find a direct association between OME or associated hearing loss and children’s later language skills. Recent randomized studies by Paradise et al3 and Rovers et al,4 in which children received either prompt or delayed insertion of tympanostomy tubes to drain middle ear fluid, as well as recent prospective correlational studies,1,2 failed to find associations between OME and language skills for children younger than those in the current study. Likewise, no association between OME and language skills has been reported by other researchers12–22 who work with children of the same ages as those in the current study. Gravel and Wallace17 found no association in high-risk and low birth weight infants between previous OME history and receptive or expressive communication at 4 years of age. Roberts et al,18 in a different cohort of low- and middle-income children attending child care, also found that receptive and expressive language for children aged 4.5 to 6 years was not related to previous OME history. Other studies examining specific aspects of language in children aged 4 to 8 years, including syntax13,14,16,19,21 and vocabulary,12,15,19,21 reported no association between OME and language.
In regard to academic achievement in the early elementary school grades, we found in this study a significant relationship between early OME and hearing loss and lower math scores on a verbal math problems task at kindergarten entry, but the relationship was no longer significant in the early elementary school grades. Thus, children caught up in their math skills once they entered school. These findings are consistent with our previous report30 of an association at kindergarten entry between both OME and hearing loss and math skills. The Applied Problems subtest measures a child’s skill in analyzing and solving simple verbal math problems. It involves the child’s determining which parts of the verbally presented information are relevant for the calculation and requires working memory and the processing of auditory-based information. It seems that, although these skills may be initially vulnerable to a history of OME or hearing loss (or possibly to some other factor related to both OME/hearing history and these math skills), by the time children are in second grade, a history of OME or hearing loss is no longer contributing to the development of these math skills.
We also found in this study no relationship from kindergarten entry through second grade between children’s OME history and letter-word identification, a component of reading, or in the identification of incomplete words, an auditory closure task. Although we had found a significant finding for auditory closure at kindergarten entry in our previous report,30 our longitudinal analyses through second grade did not support that children with more hearing loss associated with OME have lower auditory closure scores. Thus, longitudinal analyses are essential in determining patterns of growth in development and understanding OME developmental linkages.
A small number of previous prospective studies examining the relationship between OME history and children’s early reading and math skills have had differing results. Teele et al14 found that children who experienced more OME in the first year of life scored lower on achievement tests of math and reading at 7 years of age, even when factors such as socioeconomic status and gender were controlled. Gravel and Wallace17 reported lower reading scores for 6-year-old children who had a history of OME and had been high-risk and low birth weight infants. Lous et al20 also reported lower reading scores for a cohort of low- and middle-income children from Denmark. However, Roberts and colleagues23,24 reported, in a different cohort of children, no associations between early OME history and later math and reading skills (measured at kindergarten entry and, for low-income children who attended child care, in second grade). Furthermore, Peters et al16 found no association between early OME history and later reading or math skills at 7 years of age, although children with more OME were reported to score lower in writing and spelling. Johnson et al,22 in a study of 294 children, did not find a relationship between OME history and achievement scores at 5 years of age. Many of the previous studies failed to consider the role of important child and environmental factors that affect children’s early school readiness skills and thus are hard to compare with the current study.
As reported in our previous study,30 the home environment was consistently and significantly a predictor of children’s language development and academic achievement. It has been well-documented that dimensions of the home environment predict children’s language development and early academic skills.25–27 The importance of the home environment was highlighted in the figures from this study showing that from 4 years of age through second grade, the stronger association is between language development and family factors rather than between development and OME or hearing loss. Factors such as the amount of talk and style of parents’ conversational interactions (eg, responsive and elaborative); reading to a child (by describing and focusing on meanings and inferences); and overall responsiveness, organization, and support of the home environment have also been shown to predict children’s language development (during the early preschool years and early elementary school years).25–27
This study had many strengths as compared with previous studies. It is unique in that it used longitudinal analyses to examine OME association with development at 1 age point, as well as over time, to determine whether the relative strength of any association changes as the child gets older. Few studies have used longitudinal analyses to examine the linkage of OME to development. Second, the study was prospective, following a group of children who attended child care from infancy through the early elementary school years and using a standard protocol to assess OME status, hearing, and children’s development at repeated intervals. Third, children’s hearing sensitivity and OME experiences were measured frequently, when children were both well and ill with OME. Hearing, rather than OME, is believed to be the critical variable in understanding the potential linkage of OME to development. Finally, family background variables (eg, responsiveness and supportiveness of a child’s home environment), known to be important predictors of children’s development, were assessed and controlled for in the analyses.
Study results should be interpreted with caution for several reasons. First, all children are black, were recruited from community child care programs, are primarily from low-income families, and had a high incidence of OME, as described in our previous work.28–30,36,37 Thus, characteristics of this study population may restrict generalization to other study populations. Second, we did not examine specific measures of children’s language development (eg, syntax) or aspects of speech, attention, auditory processing, and behavior, choosing to use instead more global measures of receptive and expressive language. Analyses of other measures of language as well as academic achievement as outcomes may lead to different results. Third, the significant associations between OME/hearing loss and the expressive language and math scores as well as between the home environment and the language and academic outcomes were mild to moderate in degree and suggest that multiple factors in addition to the variables studied likely influence children’s development. Finally, because the analyses were correlational, the results do not suggest that OME or hearing loss causes later delays in expressive language or math skills.
There was not a significant relationship between a history of OME or hearing loss and children’s later academic skills in reading or word recognition during the preschool and early elementary school years. Although children with more OME and hearing loss scored lower in expressive language and math at the younger ages, they caught up in math on entering school and in expressive language by second grade. Children’s home environment was more strongly related to language and academic outcomes than was OME or hearing loss and continued to be predictive of language and academic outcomes through second grade. Study results should be interpreted cautiously when generalizing to other populations because the children began attending child care in infancy, had OME in early childhood, and were primarily from low-income families. Additional analyses are examining the linkage of OME and hearing loss to specific domains of language and auditory functioning and to the interrelationships among children’s OME and hearing history, family and classroom environments, and children’s language and academic skills.
This research was supported in part by the National Institute on Deafness and Other Communication Disorders (01R01-CD03817-01A1), the Maternal and Child Health Program (MCJ-370599, MCJ-370649, MCJ-00145 Title V, Social Security Act), and Health Resources and Services Administration, US Department of Health and Human Services.
We thank the children and families who participated in this study. We have greatly appreciated the support and assistance of Dr Judith Gravel, Dr Ina Wallace, Dr Sandra Jackson, Dr Eloise Neebe, Dr Jackson Roush, Dr Martha Mundy, and Sarah Henderson in this work.
- ↵Feldman HM, Dollaghan CA, Campbell TF, et al. Parent-reported language and communication skills at one and two years of age in relation to otitis media in the first two years of life. Pediatrics.1999;104(4) . Available at: http://www.pediatrics.org/cgi/content/full/104/4/e52
- ↵Paradise JL, Dollaghan CA, Campbell TF, et al. Language, speech sound production, and cognition in three-year-old children in relation to otitis media in the first three years of life. Pediatrics.2000;105 :1119– 1130
- ↵Rovers MM, Straatman H, Ingels K, van der Wilt GJ, van den Broek P, Zielhuis GA. The effect of ventilation tubes on language development in infants with otitis media with effusion: a randomized trial. Pediatrics.2000;106(3) . Available at: http://www.pediatrics.org/cgi/content/full/106/3/e42
- Roberts JE, Wallace IF. Language and otitis media. In: Roberts JE, Wallace IF, Henderson F, eds. Otitis Media in Young Children: Medical, Developmental and Educational Considerations. Baltimore, MD: Paul H. Brookes Publishing Co, Inc; 1997: 133–161
- ↵Ruben RJ, Haggard MP, Bagger-Sjoback D, et al. Complications and sequelae. Recent Advances in Otitis Media Report of the Sixth Research Conference. 1998; 174(suppl): 81–94
- ↵Stool SE, Berg AO, Berman S, et al. Managing Otitis Media With Effusion in Young Children: Clinical Practice Guidelines, Number 12. Rockville, MD: Agency for Health Care Policy and Research, Public Health Service, US Department of Health and Human Services; 1992: 1–18. AHCPR Publ. No. 94-0622
- ↵Schappert SM. Office visits for otitis media: United States, 1975–90. National Center for Health Care Statistics. Advance Data.1992:214
- ↵Teele DW, Klein JO, Chase C, et al. Otitis media in infancy and intellectual ability, school achievement, speech, and language at age 7 years. J Infect Dis.1990;162 :685– 694
- ↵Feagans L, Sanyal M, Henderson F, Collier A, Applebaum M. Relationship of middle ear diseases in early childhood to later narrative and attention skills. J Pediatr Psychol.1987;12 :581– 594
- ↵Teele DW, Klein JO, Rosner BA, The Greater Boston Otitis Media Study Group. Otitis media with effusion during the first three years of life and development of speech and language. Pediatrics.1984;74 :282– 287
- ↵Peters SAF, Grievink EM, van Bon WHJ, Schilder AGM. The effects of early bilateral otitis media with effusion on educational attainment: a prospective cohort study. J Learn Disabil.1994;27 :111– 121
- ↵Lous J, Fiellau-Nikolajsen M, Jeppesen AL. Secretory otitis media and language development: a six-year follow-up study with case-control. Int J Pediatr Otorhinolaryngol.1998;15 :185– 203
- ↵Johnson DL, Swank PR, Owen MJ, Baldwin CD, Howie VM, McCormick DP. Effects of early middle ear effusion on child intelligence at three, five, and seven years of age. J Pediatr Psychol.2000;25 :5– 13
- ↵Roberts JE, Sanyal MA, Burchinal MR, Collier AM, Ramey CT, Henderson FW. Otitis media in early childhood and its relationship to later verbal and academic performance. Pediatrics.1986;78 :423– 430
- ↵Roberts JE, Burchinal MR, Collier AM, et al. Otitis media in early childhood and cognitive, academic, and classroom performance in the school-aged child. Pediatrics.1989;83 :773– 786
- Hart B, Risley TR, eds. Meaningful Differences in the Everyday Experience of Young American Children. Baltimore, MD: Paul H. Brookes Publishing Co, Inc; 1995
- ↵Roberts JE, Burchinal MR, Zeisel SA, et al. Otitis media, the caregiving environment, and language and cognitive outcomes at two years. Pediatrics.1998;102 :346– 352
- ↵Roberts JE, Burchinal MR, Jackson SC, et al. Otitis media in childhood in relation to preschool language and school readiness skills among black children. Pediatrics.2000;106 :725– 735
- ↵Wiig E, Secord WA, Semel E. Clinical Evaluation of Language Fundamentals-Preschool. San Antonio, TX: Psychological Corporation, Harcourt Brace Company; 1992
- ↵Semel E, Wiig E, Secord WA. Clinical Evaluation of Language Fundamentals. 3rd ed. San Antonio, TX: Psychological Corporation, Harcourt Brace Company; 1995
- ↵Woodcock RW, Johnson MB. Woodcock-Johnson Psycho-Educational Battery-Revised. Allen, TX: CLM; 1990
- ↵Caldwell BM, Bradley RH. Home Observation for Measurement of the Environment Manual. Little Rock, AR: University of Arkansas; 1978
- ↵Zeisel SA, Roberts JE, Neebe EC, et al. A longitudinal study of otitis media with effusion among 2- to 5-year old African American children in child care. Pediatrics.1999;103 :15– 19
- Copyright © 2002 by the American Academy of Pediatrics