Reading Proficiency Trends Following Newborn Hearing Screening Implementation

In this study, we examine reading proficiency based on the Colorado Student Assessment Program (CSAP) and the Transitional Colorado Assessment Program (TCAP)¹³  for children with hearing loss enrolled in a large urban school district representing ≈15% of this Colorado population. Participants were students in third to 10th grades who were born between 1990 and 2006, were identified with hearing loss by educational audiologists, were enrolled in the district at some point during the 2000–2001 through 2013–2014 academic years, and had at least 1 state assessment. These years were chosen to capture children born before and after UNHS implementation.

Participants reflected the following 3 subsets of children with hearing loss based on Colorado Department of Education¹⁴  criteria: (1) bilateral hearing loss as a 3-frequency pure tone average greater than 20 dB, (2) high-frequency hearing loss as a 2-frequency average in the better ear from 2000 to 6000 Hz, and (3) unilateral hearing loss as a 3-frequency pure tone average of ≥35 dB not reversible within a reasonable period of time. Children with additional disabilities in which the other disability was judged primary by an individual educational plan were not included. Children in a grade lower than third grade and older children with language and/or reading proficiency far below third-grade levels were exempt from testing. Per district policy, children in non–English-speaking homes who had not resided in Colorado for 3 years were similarly exempt. The number of children with hearing loss taking the CSAP and TCAP increased each year after UNHS implementation, reflecting fewer exemptions due to low language levels.

Three hundred thirty-seven children with a total of 1475 assessments met these criteria, reflecting all students with hearing loss in the district as defined above. Fifty-three assessments were missing data on at least 1 non–hearing-related predictor; all 53 were missing additional disability services (DISAB+); 44 were also missing English not spoken in the home (ENSH) status, and 44 were also missing FRL status. This resulted in a final sample of 1422 assessments based on 321 unique children. The number of assessments each academic year for children from each of the birth year cohorts is summarized in Table 1. Before UNHS and in the study’s early years, most children with bilateral hearing loss were exempted from taking the CSAP and TCAP because language levels were not sufficient. A district administrator estimated that ≥150 students with hearing loss were exempted from testing in 2000, but few have been exempted since UNHS and EHDI implementation.

Reading proficiency is based on grade-specific criteria on the annual statewide standardized assessment. Four ordinal-scale proficiency levels are provided: unsatisfactory, partially proficient, proficient, and advanced. Given the low number of advanced scores (n = 23 across 1422 assessments), proficient and advanced levels were combined into 1 group, resulting in a 3-level ordinal scale.

Birth year was estimated on the basis of age reported in school data. Therefore, birth year reflected the academic year cycle and was centered on the 2000–2001 academic year (eg, births during the 1991–1992 academic year were coded 9, whereas births during the 2005–2006 academic year were coded 5).

Grade is a student’s grade level in school during a given academic year.

DISAB+ indicates whether a child received non–hearing-related special education services (0 = special education for hearing loss only; 1 = non–hearing-related special education).

FRL status was included as a marker for family socioeconomic status (0 = not eligible; 1 = eligible for FRL).

ENSH was coded as a dichotomous variable (0 = English, 1 = language other than English).

Laterality and degree of hearing loss was coded into a 3-level categorical variable: (1) unilateral, irrespective of degree; (2) mild-moderate bilateral; and (3) moderate-severe to profound bilateral hearing loss. Unilateral hearing loss served as the referent group.

Because children born in different years experienced different exposure to UNHS and EHDI services, the effect for time was divided into 2 variables. Birth year was the year in which the child was born; later birth years reflected increased exposure to UNHS and EHDI diagnostic and early intervention. Thus, birth year captures the between-child differences based on birth cohort. Grade in school reflected the annual growth (or decline) in proficiency a child experienced, capturing the within-child differences over time.

Analyses began with multilevel (assessments nested within students) logistic regressions examining trends in the child characteristics over time. The primary analyses involved multilevel ordinal logistic regressions examining birth year and grade in school as predictors of proficiency level to capture the overall trend over time across birth cohorts (ie, the birth year cohort effects), controlling for maturational trends within each cohort (ie, the grade level effect). These analyses also controlled for child demographic predictors (FRL, ENSH, and DISAB+) and examined possible interactions between birth year cohort and child characteristics as predictors of proficiency levels. A final exploratory analysis examined the impact of laterality and degree of hearing loss on temporal trends in proficiency. Analyses were performed by using SPSS version 26.0 (IBM SPSS Statistics, IBM Corporation)¹⁵  and HLM 8.0.¹⁶ 

The study was ruled exempt by the University of Colorado Boulder Institutional Review Board.

Descriptive statistics are summarized in Table 2. Reflecting the multilevel data, these are reported separately for level 1 data (ie, annual school data that change over time) and level 2 data (ie, child characteristics that do not change over time).

Among the 321 children, 32 (10.0%) were identified as DISAB+, whereas 85 children (26.5%) were in the ENSH group. Laterality was available for 75.4% of children, with 31.4% of those identified as having unilateral hearing loss, 42.1% as having mild-moderate bilateral hearing loss, and 26.4% as having moderate-severe to profound bilateral hearing loss.

The 1422 assessments took place across third through 10th grades, with more assessments in the lower grades. On average, children had 4.43 assessments, with a median of 4.0 and an interquartile range of 2 to 7. In 1043 (73.3%) assessments, the student was eligible for FRL, with 231 of the 321 students (72.0%) eligible for FRL at the time of at least 1 of their assessments. Reading proficiency was classified as unsatisfactory in 45.6% of the 1422 assessments (n = 649), partially proficient in 26.2% of assessments (n = 372), and proficient or advanced in 28.2% of assessments (n = 378 and n = 23, respectively). This reflected 150 children (46.7%) having at least 1 assessment classified as unsatisfactory and 94 (29.3%) having at least 1 classified as proficient or advanced (Fig 2). Preliminary analyses revealed that none of the child characteristics, including degree and laterality, varied over time.

As shown in Table 3, the number of assessments increased each year because of a new birth cohort entering the grades 3 through 10 testing window and fewer children being exempted from the test each year. Note that from 2007 to 2014, the percentage of district students without hearing loss who were identified as proficient or higher increased from 43% to 54% (odds ratio [OR] = 1.60), versus 19% to 37% for students with hearing loss (OR = 2.49).

Results of the multilevel ordinal logistic regression are presented in the first panel of Table 4. Grade and FRL status were entered as level 1 variables; all other predictors were level 2 variables, with the intercept treated as a random effect. Birth year was statistically significant even after controlling for other characteristics. Specifically, the odds of being identified with a higher level of proficiency increased by 23.3% for children born in each subsequent birth year cohort. Furthermore, the odds of being in a higher proficiency level increased by 22.0% with each year in school. Additional student factors were related to decreased odds of being in a higher proficiency level. Specifically, the odds of being in a higher proficiency level decreased by 73.9% with DISAB+, by 78.4% with ENSH, and by 71.1% with FRL status in any given year.

Follow-up analyses revealed no significant interaction between birth year cohort and DISAB+ (OR = 0.888 [95% confidence interval (CI) 0.656–1.203]), ENSH (OR = 1.106 [95% CI 0.914–1.339]), or grade in school (OR = 1.015 [95% CI 0.991–1.040]). Despite the surprising nonsignificant interactions, Table 5 (DISAB+) and Table 6 (ENSH) are included to highlight the similarity in development gain. In contrast, a significant interaction was observed between FRL status and birth year cohort (OR = 0.815 [95% CI 0.707–0.939]). Specifically, whereas a birth year cohort effect was observed among children identified as eligible for FRL (OR = 1.166 [95% CI 1.058–1.285]), the effect was greater for those not identified as eligible for FRL (OR = 1.431 [95% CI 1.244–1.646]). Although reflecting a disparity in benefits, for context, even among children eligible for FRL the odds of being in a higher proficiency level more than quadrupled (OR = 4.645) over a 10-year period (eg, children born in 2005 versus those born in 1995). This is on par, albeit it in the opposite direction, with the effects of FRL status, DISAB+, and ENSH. This differential effect is illustrated in Fig 3, which presents the annual frequency distributions of proficiency levels separately for each group.

A final set of analyses added a 3-level variable indicating degree and laterality: (1) moderate-severe to profound unilateral, (2) mild-moderate bilateral, and (3) moderate-severe to profound bilateral hearing loss, with unilateral hearing loss (ie, group 1) serving as the referent group. A total of 79 of the 321 children (24.6%) were missing this variable, and further analysis suggested data were not missing at random. For example, those missing data on degree and laterality were at greater odds of being identified as eligible for FRL (⁠$χ12$ [n = 321] = 5.527; P = .019; OR = 2.126). In addition, reflecting state criteria, children with unilateral hearing loss (<35 dB) were not classified as having hearing loss and thus not included in these data. Therefore, this analysis is seen as exploratory.

Results after adding degree and laterality are presented in the second panel of Table 4. Degree and laterality was related to proficiency level (⁠$χ22$ [n = 321] = 12.405; P = .002). No difference was found between unilateral and mild-moderate bilateral hearing loss (OR = 0.651 [95% CI 0.262–1.616]); however, the moderate-severe to profound bilateral group did have significantly lower odds of being in a higher proficiency level than the unilateral group (OR = 0.162 [95% CI 0.057–0.461]). Follow-up analyses not included in Table 4 revealed that those with moderate-severe to profound bilateral hearing loss also had lower odds of being in a higher proficiency level those with mild-moderate bilateral hearing loss (OR = 0.249 [95% CI 0.091–0.683]). Subsequent analysis found that degree and laterality did not interact with birth year cohort (⁠$χ22$ [n = 321] = 0.369; P > .500), indicating consistency across birth years.

After the implementation of UNHS and EHDI in Colorado, significant improvements in reading proficiency of successive birth cohorts of students with hearing loss were observed. Decreased exemptions from testing were associated with increases in the number of children taking state assessments each year. With each successive test year, the proportion of children in the proficient or advanced category increased as the proportion in the unsatisfactory category decreased. Gains were observed both within students, year-to-year over time, and between students in different birth year cohorts. These gains indicate that students with hearing loss made significant progress in closing the gap in reading proficiency relative to typically developing youth during the study period.¹⁷  This study is the first to demonstrate a long-term temporal improvement in reading proficiency after UNHS implementation when 80% of the Colorado population met national benchmarks. A previous UK study reported that improved reading proficiency in a UNHS cohort observed at ages 6 to 10 was no longer significant at ages 13 to 19, although teenagers confirmed with hearing loss by age 9 months had significantly better reading scores, highlighting the importance of UNHS accompanied by early identification.¹² 

Despite sociodemographic disparities, significant shifts to higher levels of proficiency since 2000–2001 were observed in all groups with hearing loss. Reading proficiency gains were similar irrespective of the language spoken in the home. Children with unilateral hearing loss had similar proficiency gains to children with mild-moderate bilateral hearing loss. The findings are consistent with previous reports from Colorado that earlier identification and intervention significantly improved the developmental outcomes of all subgroups of younger children (through 3 years of age) with hearing loss, irrespective of degree of hearing loss or presence or absence of additional disabilities and maternal level of education.^3,7,8 

Not all groups of children, however, experienced equivalent gains in reading proficiency. Children with moderate-severe to profound hearing loss had significant gains of lesser magnitude than those with unilateral or mild-moderate bilateral hearing loss. Most concerning was that although children eligible for FRL made significant gains, these gains were less than half the gains experienced by children from higher-income families. By 2013–2014, 70.5% of the children ineligible for FRL were proficient or advanced and only 4.5% had unsatisfactory scores, whereas among those eligible for FRL, 24.5% were proficient or advanced and 41.3% had unsatisfactory scores. FRL eligibility remains a powerful negative predictor of reading outcomes. These findings are consistent with literature on the relationship between disadvantage and language, cognitive, and neurologic development.^18,19 

A limitation of this study is the inability to individually link children to UNHS results, making it impossible to differentiate children with late identification and acquired and progressive losses. However, these groups should be equally represented in each test year. Another limitation is that we only had data for FRL eligibility and not the socioeconomic variables that predict FRL eligibility. Also, data are based on one district in a single state; however, this district does reflect a population that is typically seen as being at relatively higher overall levels of risk. In addition, this is not a randomized control study, but it is a population-based description of change over time. For >30 years minimal improvement in average reading proficiency of children with hearing loss^1,2  was documented until the implementation of UNHS. Efficacy of interventions, such as hearing aids and cochlear implants, has been dependent on earlier access, possible only after UNHS implementation.²⁰ 

This study provides pediatricians with further evidence that children with hearing loss post UNHS and EHDI implementation can experience substantial literacy gains. Pediatricians can encourage parents of children with unilateral or bilateral hearing loss to follow through with all audiology and early intervention appointments, which can help ensure children receive appropriate services and achieve optimal developmental outcomes. Ineligibility of children with unilateral hearing loss for early intervention services until a delay is found should be reconsidered. The children with unilateral and mild-moderate bilateral hearing loss in this study had similar developmental needs.

Of particular encouragement is the narrowed gaps in language performance between children with hearing loss relative to typically developing students. The closing of the gap in reading comprehension in a population of children with significant risk factors is an important and novel finding.

Although most groups experienced similar reading proficiency gains over time, children eligible for FRL and those with moderate-severe to profound bilateral hearing loss experienced significantly smaller gains. The increased sociodemographic disparity in reading proficiency among children with hearing loss in this study indicates a health equity issue. This is of particular concern because it suggests that the provision of intervention services after early identification may favor children in relatively advantaged families compared with those in disadvantaged families. In future studies, researchers may assess whether greater intensity or different methods of delivering interventions before third grade to children in families of lower sociodemographic status and those with moderate-severe to profound bilateral hearing loss may help to attenuate disparities in reading proficiency.

UNHS is only the first step in the EHDI process; improved outcomes depend on early diagnosis and intervention along with minimizing disparities in receipt of services.¹⁰  Children with hearing loss can benefit from multiple interventions, both within the health domain and within the educational domain. All types of intervention, whether educational or involving assistive technology, can be more effective in improving language outcomes with earlier initiation.

CI

confidence interval

CSAP

Colorado Student Assessment Program

DISAB+

additional disability services

EHDI

Early Hearing Detection and Intervention

ENSH

English not spoken in the home

FRL

free and reduced lunch

OR

odds ratio

TCAP

Transitional Colorado Assessment Program

UNHS

universal newborn hearing screening