Variables available for analysis included demographic fields (i.e., maternal age, race and ethnicity, education level of parents, and insurance status/payer). Variables that characterized the birth included gestational age, birthweight, Apgar score, hearing loss-specific risk indicators (e.g., family history, specific physical findings, neonatal intensive care stay) and pregnancy-related risk factors (e.g., in utero infections). Risk indicators were according to the Joint Committee of Infant Hearing Position Statement.¹ Hearing-specific information included age at screening and diagnosis, laterality of hearing loss (unilateral/bilateral), and degree of hearing loss (e.g., mild, moderate, moderate-severe, severe and profound). Early intervention data used in the current study included date of EI enrollment and documented developmental delays/disabilities. Presence of disabilities within the EI record was according to a form completed by medical professionals licensed to diagnose and treat mental and physical conditions believed to likely lead to developmental delays. Information on amplification was not available in the record.

All academic outcomes were part of Ohio’s assessment standards. Between 2011 and 2014, emergent literacy was measured during preschool using the Get it! Got it! Go! (GGG) assessment.²² The GGG is composed of three individual indicators in children’s language and literacy development: Picture Naming (measuring expressive language development), Rhyming, and Alliteration (measures of phonological awareness associated with early literacy development).²³ Each task is individually administered and timed. Picture naming (one minute) requires the child to quickly name pictures of familiar objects found in home, classroom, or community settings. Rhyming (one minute) requires the child to point to one of three pictures that rhymes with a target picture, after all pictures are named by the examiner. Alliteration (two minutes) requires the child to identify one of three pictures that starts with the same sound as the target picture. GGG scores reflect the number of items answered correctly in the time limit; no maximum or standard score exists. These three indicators have good test-retest reliability (from r=0.67 to 0.89), significantly correlated with other language standardized measures that are a correlate of literacy, are sensitive to change,²³ and have been used in previous emergent literacy research.²⁴ The GGG was administered in the Fall and Spring of each school year. The number of available assessments varied (918 observations total). Fifty children had one GGG assessment date, 65 had two assessments and the remainder had three or more (range 1–10; mean 3.3).

The initial Individualized Family Service Plan date was used as the EI enrollment date. Early intervention exposure was defined using the national EHDI benchmark of enrollment into EI before age 6 months. Children were classified as entering EI early if they enrolled before age 6 months and as entering EI later if they enrolled at/after age 6 months.

Statistical analyses were conducted using SAS version 9.4 (SAS Institute Inc., Cary, NC) software. Data distributions were assessed for normality. Socio-demographic and clinical variables were tested between EI early group and EI late group using Pearson chi-square or Student’s t test as appropriate. In order to balance the characteristics of children who entered EI early vs. later, we employed a propensity score (PS) matching. PS is the probability of group assignment based on observed baseline characteristics and usually obtained using logistic regression where group label (EI early vs EI late) will be modeled as a function of baseline variables. Variables included in the logistic regression model are: gender, birth weight, gestational age, race, maternal education level, insurance status, degree of hearing loss, presence of risk indicator for hearing loss, and presence of disability. Once the PS was generated, we used nearest-neighbor matching approach in which each subject in the EI late group is sequentially matched with one or more (up to 5) subjects in the EI early group, with a caliper width of 0.25 of the logit of the propensity score (i.e. subjects are considered for matching if the difference in the log propensity score is <0.25).

The relationship of emergent literacy skills development (measured in preschool) over time (age of child) as a function EI enrollment (EI early vs EI later) was investigated using mixed effects models (PROC MIXED) with a random intercept to account for the matched design and a repeated statement with a compound symmetry covariance structure to account for potential correlation induced by repeated data measures. Three models were constructed for the GGG domains of picture naming, rhyming, and alliteration. Variables that remained significantly different between the EI enrollment group after matching were included in the model (high risk indicator of hearing loss, disability diagnosis, bilateral vs. unilateral hearing loss). A group by age interaction term was tested in each model to examine whether trajectories across time differed by group. Estimated trajectories (based on least square means) of emergent literacy outcomes derived from these models were graphically illustrated. Outcome trajectories of all Ohio preschoolers during the same time frame were illustrated as a reference. Because we did not have the same detailed data on all Ohio preschoolers, we did not statistically compare the trajectories of DHH children with all preschoolers.

Domain scores that were classified as “not reported” were coded missing (accounting for 6.8% picture naming, 12.1% rhyming, 10.6% alliteration). To reduce bias and maximize use of available information, we conducted a multiple imputation procedure to impute missing data along an estimate of uncertainty about the imputed data. We used a regression approach, including relevant covariates to impute missing data, creating a complete dataset. We repeated this process 20 times, generating 20 complete datasets. We then fit the mixed model described above and generated estimated parameter values for each dataset. We pooled the parameter estimates (including standard errors) from each dataset to obtain a combined estimate which were reported as results of the final model. Finally, we performed sensitivity analyses to understand potential biases in our estimates given missing outcome data. First, we rerun our models treating the not reported scores as “0”, (instead of setting them as missing) possibly biasing our estimated slopes towards zero. We also reran the models using only available data (without imputation).

Within the Ohio EHDI Data Linkage Project, 1262 born between 2008 and 2014 were enrolled into EI. The GGG was administered to 276 children enrolled in preschool during the school years 2011–2014. Thirteen additional children had been enrolled in preschool by 2014 but had no GGG assessment data available (10/13 were enrolled in 2014). Of the 276, 160 (58%) had been enrolled in EI before age 6 months. Propensity score matching resulted in 102 matches for 256 children. Twelve children had missing match data and eight children did not match at all. Of the 256 matched children, 154 (60.2%) had enrolled in EI before age 6 months.

Children included in the current analysis looked similar to all DHH children enrolled in EI (Table 1). Compared to children who entered late, those who entered early were more likely (p<0.05) to have had hearing loss confirmed early (median 2.9 months vs. 8.4 months), a risk indicator for hearing loss (56.5% vs 35.3%), a disability diagnosis reported in the EI system (44.8% vs 28.4%), and less likely to have bilateral hearing loss (82.5% vs. 93.1%) (Table 1).

Model results indicated that the group by age interaction term was not statistically significant for all three models. DHH children who enrolled in EI early had significantly higher mean scores (mean, 95% CI) over time compared to children who enrolled later for the GGG domains of picture naming (2.42, (0.47, 4.37)), approximately 2.4-word difference), rhyming (1.2, (0.35, 2.06)), approximately 1-word difference) and alliteration (0.61, (0.15, 1.07)). (see Table 2 for unadjusted and adjusted model results). All models accounted for the pre-specified set of covariates listed earlier, including assessment age in years, presence of disability, presence of risk factor for hearing loss, and laterality (unilateral vs. bilateral). Figure 1a–c illustrates the trajectories for the GGG domains of picture naming, rhyming, and alliteration for DHH children and for Ohio preschoolers as a reference. DHH children had similar trajectories for picture naming as Ohio preschoolers. Ohio preschoolers had steeper trajectories for rhyming and alliteration, indicating a faster pace of learning for these two domains. The sensitivity analyses resulted in consistent results with the final models (Supplemental Digital Content Tables 1 and 2).

For DHH infants and toddlers, Part C EI initially serves as a proactive approach supporting language development in a population that is considered high risk for language delays. There is considerable evidence supporting the impact of UNHS and EI on decreasing ages of identification and improving language and vocabulary development in DHH children.^1,4,14,15 Early ages of EI enrollment have provided stronger evidence towards improved language outcomes for this population. Although earlier hearing loss confirmation has been associated with better reading abilities at primary school age,¹⁹ little is known regarding how early enrollment into EI may impact reading abilities. It is encouraging that findings from the current study indicate that the benefit of early EI enrollment (prior to age 6 months) can be seen with emergent literacy measured in preschool.

In Ohio, DHH infants and toddlers with their families receive specialized EI services focused on goals that support the needs specific to DHH educational practice. Earlier enrollment into EI services allows for a longer period of time within EI (longer exposure time). Early EI enrollment may have provided more opportunities for EI providers to develop an effective relationship with families and coach families on early language stimulation. Transitioning from EI to academic settings shifts from a high level of direct family coaching and language focus to the structured setting focusing on academic and social performance. EI and academic systems are not always congruent, with distinct oversight agencies in some circumstances, differing funding sources, and disparate data systems in which information regarding an individual child are tracked. Our study had the benefit of strong agency partnerships and the foundation of a robust data linkage study.

Emergent literacy skills provide the foundation for later reading and academic development in all children²⁸ including in DHH children.¹² Reading comprehension in DHH children remains below that of hearing peers despite technological advances (e.g., cochlear implants) which have facilitated improvements in speech perception and production abilities. Many DHH children enter kindergarten behind their peers regarding literacy and this gap persists into school age; a high proportion of students have consistently had reading skills below age-appropriate or grade-level.^3,6,29 Previous emergent literacy studies in DHH children have not evaluated the impact of EI exposure in the first 36 months of life on these outcomes. Our study provides a different approach to understanding emergent literacy by emphasizing the importance of the early receipt of EI on these later outcomes. Children who received EI before age 6 months had consistently higher emergent literacy outcomes in picture naming, rhyming, and alliteration than children who received EI at/after age 6 months. A striking finding was the inability for the later EI group to close the gap over time; they entered preschool with gaps in emergent literacy skills compared to early EI children and these gaps remained over time. We believe it is possible that providing children with a strong foundation for language development as early as possible (prior to 6 months in this case) indirectly provides them with necessary tools for emergent literacy.

It is important to note that DHH children did not always appear to enter preschool with emergent literacy scores lower than the Ohio preschool reference group. Because our data were limited regarding the reference group, we were unable to determine levels of ability for this group. It is likely that some preschoolers could have language or cognitive disabilities that would affect their academic progress. We were unable to control for such factors in the broader population as compared to the models for DHH children. Although our primary goal was to compare the EI early vs. EI later groups of DHH children, and not directly compare DHH children to children with hearing, the difference in trajectories for rhyming and alliteration warrant further investigation. The persistent lower scores on the GGG domains for DHH preschoolers who enrolled into EI later (at/after age 6 months) emphasize the potential need for additional support early on regarding emergent literacy. The differences in scores between early and later EI groups on these limited timed assessments were not only statistically significant, but also academically meaningful. It has been shown that it is difficult for children to “catch up” to their peers in formal academic settings when they start with fewer skills, even if the skill gap is narrow.³⁰ Due to the greater risks for language-based skill delays, seemingly small differences in pre-literacy skills during preschool for DHH children allow for early recognition of needs for targeted intervention. Our findings support a positive association with early EI enrollment on continued development of these emergent literacy skills. Emergent literacy warrants further attention for DHH preschoolers.

Study strengths include a large representative sample size of young Ohio -born children identified with hearing loss, and repeated outcome measures. Data were collected in the same way for all children and were linked across different state agencies providing a unique opportunity to understand outcomes occurring beyond the EI period. This has allowed us to evaluate a large longitudinal cohort from birth into school ages. We acknowledge limitations of this research. Data were collected through public health and education agencies. Although the potential for unmeasured confounding exists, we conducted a matched analysis to minimize potential bias and confounding that may exist between the groups. Although there was imbalance with regard to a few of the variables (disability, risk indicator, laterality), these factors were more prevalent among children who enrolled into EI early. These factors also did not appear to confound the relationship between EI enrollment and outcome, according to the regression models. We did not have current language skills for preschoolers, which could help elucidate emerging literacy levels. This study was not designed to understand the mechanisms of skill development, but to assess the association of EI with later skills. Audiologic data was only available at EI enrollment; we were unable to account for current hearing levels. Reliance on specific technology (i.e. hearing aids, cochlear implants) and aided responses were not reported. Although device information may be important for understanding factors associated with outcomes, we do not believe that it would impact the relationship between age of EI (early vs. later) and outcomes. We included data on whether a child had a diagnosed disability that may be related to high risk for developmental delays, though we had no specific information on cognitive development. At the time of this study, the EI record only included whether a developmental assessment was scored below 2 standard deviations of the population for any developmental domain. Since no actual score was provided, and the record had no indication of when scores were above the cutoff of 2 standard deviations, we were not confident in using this variable to indicate cognitive development. The inclusion of the variable indicating the presence of disability appeared to be a robust predictor of outcomes. Future research should include more specific information on cognitive development. Education data from the state department include children on Individualized Education Programs (IEP), regardless of setting. A child in a private educational setting without an IEP would not have available data. Finally, the assessments available did not provide standard scores. The GGG is a standardized evaluation that targets key early literacy components to allow educational systems to track a child’s trajectory over time. Since rhyming and alliteration have been noted to be areas of challenge for DHH children,³¹ the GGG allowed us to assess these areas in the context of EI exposure. We were able to compare GGG data of DHH children with all Ohio preschoolers, allowing for a reference comparison.

The current findings extend previous research demonstrating the continued benefits of receiving early intervention before age 6 months, an important EHDI benchmark. While much research has focused on EI benefits on language outcomes during the first 36 months of life (the EI period), we focused on novel outcomes of preschool emergent literacy skills, which require a language foundation. Child outcomes beyond the early intervention period are important to track and understand. Understanding the association between these early childhood investments and the later developmental needs that occur during early school ages help provide further evidence of the long term importance of EI. By evaluating components of emergent literacy, we not only see that children who are enrolled in EI before 6 months of age have consistently higher scores in the areas of picture naming, rhyming, and alliteration over time compared to children enrolled at/after 6 months of age, it appeared that children enrolled early did as well as the preschool reference group at preschool enrollment and with picture naming over time. These findings are important because they highlight an area in which EI interventionists can help facilitate improved outcomes in the preschool/kindergarten transition. Although there have been interventions to improve reading in DHH children,³² research is needed to understand intervention strategies in first 36 months of life designed to enhance later reading abilities during for DHH children. Though not a current focus in EI, considerations for emergent literacy should be discussed, particularly for DHH children. Early EI enrollment should be a high priority for children who are identified with a hearing loss.