A Nationwide, Population-Based Study of Intelligence and Hearing Loss among 3,104,670 Adolescents

Hearing impairment is one of the most prevalent medical conditions. In 2018, 6.1% (466 million) of the world’s population was affected, 34 million (7%) of whom were children [1]. The World Health Organization (WHO) projects that the prevalence of hearing impairment will reach 933 million in 2050 [2]. In the USA, nearly 1 in 4 individuals aged 12 years or older have hearing loss (HL) in at least one ear, and 1 in 7 have bilateral HL [3]. Moreover, approximately 2.2 million Americans aged 12 years or older have severe to profound HL in at least one ear [4].

HL has been correlated with various comorbidities, including vision impairments [5], balance disorders [6], diabetes mellitus [7], cancer [8], and cardiovascular [9] diseases. Moreover, in cross-sectional analyses, cognitive impairment was more prevalent in people with HL, both tested [10, 11] and self-reported HL [10]. The relation between HL and general intelligence (GI) has also been examined in many longitudinal studies. For example, several studies [10‒12] found increased risks of dementia among people with than without HL, with hazard ratios typically in the range of 1.2–1.5. In addition, a negative relationship was found between the prevalence of HL and literacy rates among children and their parents [13]. The detrimental effect of HL on communication may impair psychosocial health, including depression [14] and anxiety [15]. Other psychosocial concerns are social isolation and loneliness [16].

Sensorineural HL (SNHL) is a permanent hearing impairment caused by a cochlea or auditory nerve defect. SNHL is correlated with memory impairment, speech difficulties, and lower intelligence scores [17, 18]. Conductive HL (CHL) is characterized by reduced sound transmission through the external auditory canal and middle ear structures, while bone conduction remains normal [19].

Although several observational studies have described the relations of HL with both GI impairment and psychosocial health, only a few population-scaled studies have been published. Most of these involved the adult population. They included limited relations to measured intelligence tests and severity levels of HL and CHL, and they primarily comprised individuals with CHL and other maxilla-facial or socioeconomic concerns [20]. This study aimed to compare GI between adolescents with SNHL, CHL, and normal hearing and evaluate the correlation between HL severity and the level of GI in a large national cohort of adolescents.

This retrospective study from a large, unselected population with premilitary health examinations during late adolescence assessed the relationship between GI and HL. Our study adheres to the guidelines of the STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) for reporting observational studies. Ethical approval for this retrospective study was obtained from the IDF Ethics Committee under approval number IRB-IDF 2072-2020, in accordance with the Helsinki Declaration and its later amendments.

Israeli adolescents undergo medical and psychosocial assessments 1 year before mandatory military service, between ages 17 and 19 years (mean age 18 ± 1 year). These include the completion of a detailed questionnaire that accesses demographic and medical factors, after which physicians at enlistment centers perform complete physical examinations. Information regarding country of origin, education, and residential socioeconomic status (SES) is collected, in addition to a mandatory GI test (detailed below) [21‒23]. All medical information is recorded in the same central database, using uniform follow-up [24, 25]. Arab citizens, Druze women, and orthodox/ultraorthodox religious Jewish women are exempt from military service and essentially do not undergo this compulsory assessment.

This analysis included 3,470,197 individuals who underwent prerecruitment evaluation from 1967 to 2019, irrespective of whether they served in the army. Excluded from the analysis were 252,639 individuals for whom GI test results were unavailable and 112,707 with prior psychiatric or neurologic diagnoses. Figure 1 illustrates the study design.

Given the distribution of HL in our study, where most adolescents had normal hearing, and a smaller proportion had HL, we employed specific statistical methods to address potential challenges associated with this imbalance. We utilized multivariate logistic regression models to control for potential confounders and enhance our analysis’s robustness. The study’s large sample size ensured adequate statistical power, allowing for the detection of significant associations even within the smaller subgroup of adolescents with HL. These methods were chosen to ensure the findings are reliable and representative of the population studied.

The GI test is conducted as part of military recruitment assessments. In addition, the test has been used extensively as an investigative tool, as previously described [26, 27]. The validity of the GI test as a measure of GI was demonstrated by its correlation of over 0.8 to the Wechsler Adult Intelligence Scale total IQ [27‒30]. The GI test evaluates language skills and academic abilities utilizing four subtests;

• 1.

  Arithmetic-R: Assesses mathematical reasoning and concentration.

• 2.

  Raven’s Progressive Matrices-R: Measures abstract nonverbal reasoning and visual-spatial problem-solving capabilities.

• 3.

  Otis-R: Tests verbal comprehension and reasoning.

• 4.

  Similarities-R: Evaluates verbal abstraction and categorization [27].

The sum of the four test scores forms a validated measure of GI (IQ) [28]^, which is correlated to IQ between 70 and over 130 [27]. Although each subtest targets specific cognitive faculties, our analysis was conducted using the composite score of these subtests. Due to the data provided in aggregate form by the military recruitment assessments, we could not access individual subtest scores.

According to the military recruitment procedure, the medical history of the adolescents is obtained from family physicians using a structured protocol. When a record contains the diagnosis of HL or an affirmative reply to any question regarding aural symptoms, such as tinnitus, aural fullness, and subjective HL, the recruit is referred to an otolaryngologist for further evaluation. According to the pure tone audiometry test, HL, and the otolaryngological assessment, adolescents are assigned an HL type (SNHL, mixed HL, or CHL) and severity (Table 1). A few differences should be addressed when comparing HL, which was assessed by the military recruitment centers (MRC) and by the American National Standards Institute (ANSI) 31 criteria. Initially, it should be noted that the Military Recruitment Centers' (MRC) criteria incorporate the hearing threshold of the contralateral ear – typically the ear with better hearing – into the overall classification. This approach reflects the understanding that communicative functioning is predominantly influenced by the ear with better hearing. Second, in assessing SNHL, the MRC criteria focus on specific frequencies, primarily those necessary for essential communication. Although high-frequency HL is considered less severe in terms of overall functionality, it is crucial to recognize that these frequencies are critical for speech intelligibility, particularly for the perception of consonants, which are essential for understanding speech. Vowels, which provide the energy in speech, predominantly occupy the lower frequencies, typically up to 1,000 Hz. Third, while some HLs are mild by ANSI criteria, they are rated as moderate to severe by MRC criteria; this indicates that the military criteria are more strict and less flexible. Fourth, MRC criteria do not include a category of profound HL, in contrast to ANSI. Finally, while mixed HL was first included in our observation, this diagnosis has been used for only a few years. Thus, only 63 recruits received this specific diagnosis and were excluded from the analysis.

The Israeli Ministry of Interior and the Israeli Central Bureau of Statistics obtained SES data based on residence. Accordingly, all municipalities are stratified on a 1–10 scale. Factors considered in the determination of SES include age distribution, available workforce, level of unemployment, level of education (the proportion of undergraduate students and those entitled to a high school diploma), average income per capita, and the proportion of income support recipients. As reported previously [31, 32], SES was categorized into three groups: low (SES 1–4), medium (SES 5–7), and high (SES 8–10).

In our study, educational attainment is classified into “low” and “high” education. “High” education refers to completing 11 full years of education or more, which typically corresponds to completing high school. Conversely, “low” education is defined as less than 11 years, indicating that the individual did not complete high school.

Country of origin (classified by the father’s or grandfather’s country of birth) and country of birth were categorized into five geographical areas: former USSR countries, Asia (relating to west Asia and not southeast Asia), Africa (mainly relating to North Africa, excluding South Africa), Western (West Europe, North and South America, South Africa, Australia, and New Zealand), and Israel.

The GI test was analyzed as a categorical variable comprised of four scores, as reported previously [31]. Q1 was low GI, Q2 was low to medium GI, Q3 was medium GI and finally, Q4 was high GI.

All statistical analyses were performed using the Statistical Package for the Social Sciences, version 26.0 (SPSS, Inc., Chicago, IL, USA). Mean values, standard deviations, Student’s t test, and the one-way ANOVA test were used for continuous variables. For categorical variables, the χ² test and Kruskal-Wallis test were used. p values for all hypothesis testing were two-sided, and p < 0.05 was interpreted as statistically significant. Multivariate binary logistic regressions were used to estimate the odds ratios (ORs) and 95% confidence intervals (CIs) for HL outcomes among GI test categories. Q4, high GI was the reference category. Models were used to assess these relations after adjusting for possible confounders such as age, sex, education, and SES stepwise. First-step unadjusted analyses were done; second adjustments were made for age and sex, third for SES, and last for education and country of origin. Due to the small numbers, the statistical analyses did not include 63 (0.3%) persons with mixed HL nor multivariate binary logistic regressions of the association between level 6 HL severity of SNHL and GI categories.

Of the 3,470,197 adolescents with prerecruitment evaluation from 1967 to 2019, 365,346 were excluded due to missing GI data or prior psychiatric or neurologic diagnoses (Fig. 1). Of the 3,104,851 adolescents included in the study, 20,075 (0.06%) were diagnosed with HL. Of them, 14,659 (73%) were diagnosed with SNHL, 5,353 (27%) with CHL, and 63 (0.3%) with mixed HL. Among those with SNHL, 17%, 43%, 15%, 16%, 8%, and 1% met the hearing severity levels of 1–6, respectively. Among those with CHL, 33%, 39%, 23%, and 5% met the hearing severity levels of 1–4, respectively. The definitions and distribution of HL severity are detailed in Table 1.

Table 2 presents the baseline characteristics of the adolescents according to the four GI categories. A higher proportion of adolescents born in West Asia or Africa than in other regions had GI scores in the lowest category (Q1). GI was associated with the level of education and SES.

The distribution of GI categories remained constant during the study period, among the entire study population (p = 0.98, online suppl. Fig. 1a; for all online suppl. material, see https://doi.org/10.1159/000542157) and also among the subgroups, with SNHL (p = 0.95, online suppl. Fig. 1b) and with CHL (p = 0.25, online suppl. Fig. 1c). In sub-analyses, differences over the study period in the distributions of the GI categories among the hearing severity groups were not statistically significant; the p values were 0.6 and 0.5 for SNHL and CHL, respectively. However, throughout the study period, a higher proportion of those with CHL than with SNHL were scored in the lowest GI category.

Age, sex, level of education, country of origin, and SES were added as confounders to the multivariate analysis models for HL outcome among GI categories. For the SNHL group, the adjusted ORs for the low, low to medium, and medium GI categories relative to the highest GI category were 1.3 (95% CI 1.4–1.2), 1.2 (95% CI 1.25–1.1), and 1.1 (95% CI 1.15–0.04), respectively (Fig. 2a). The corresponding ORs among those with CHL were 1.8 (95% CI 1.9–1.6), 1.3 (95% CI 1.4–1.2), and 1.1 (95% CI 1.2–0.9), respectively (Fig. 2b). For males compared to females, the ORs were 1.6 (95% CI 1.5–1.7) and 1.3 (95% CI 1.2–1.4) among those with SNHL and CHL, respectively. Considering high SES as the reference, the ORs for the low- and medium-SES groups were 1.09 (95% CI 1.04–1.13) and 1.16 (95% CI 1.09–1.20), respectively, among those with SNHL. The ORs among CHL patients were 1.09 (95% CI 1.02–1.16) and 0.9 (95% CI 0.84–1.01) for the low and medium SES groups, respectively.

Our analysis examined the relationship between HL severity and GI scores in adolescents with CHL and SNHL, using normal hearing as the reference group. For individuals with CHL, the adjusted ORs for being in the lowest GI category (Q1) remained consistent across different severity levels, with ORs of 1.6 (95% CI 1.5–1.8), 1.6 (95% CI 1.4–1.7), 1.6 (95% CI 1.4–1.8), and 1.7 (95% CI 1.2–2.3) for HL levels 1 through 4, respectively. This consistency suggests that CHL does not exhibit a severity-dependent relationship with GI scores. Conversely, in individuals with SNHL, there was a noticeable increase in the ORs as HL severity increased, with ORs of 1.2 (95% CI 1.1–1.3), 1.1 (95% CI 1.0–1.2), 1.2 (95% CI 1.1–1.3), 1.4 (95% CI 1.2–1.5), and 1.3 (95% CI 1.1–1.5) for HL levels 1 through 5, respectively. These findings indicate a severity-related association between SNHL and lower GI scores, underscoring the more pronounced impact of SNHL severity on cognitive performance compared to CHL (Table 3).

In this cohort of 3,435,540 healthy adolescents, 20,075 (0.06%) were diagnosed with HL. An inverse association was demonstrated between GI and the incidence of HL. The risk was higher for lower GI among those diagnosed with CHL than SNHL. Furthermore, while SNHL demonstrated a severity-based decline in GI scores, CHL did not show a similar correlation. Finally, the association between HL and GI persisted after adjusting for multiple confounders, including age, education, SES, country of origin, and birth, thus supporting an independent association between hearing impairment and GI.

Several articles have prospectively evaluated the association between GI and the vast spectrum of health and survival-related outcomes [26, 32, 33]. Our study used a tool based on mathematical, verbal, and nonverbal reasoning and visual-spatial problem-solving abilities. The tool was validated in multiple studies [26, 27, 34]. Deary et al. [26] demonstrated that our tool in this work had similar results to other validated intelligence tests [27, 35].

Another strength of our manuscript is the diverse population due to the relatively high rate of immigration in Israel. Thus, as was previously concluded [36, 37], our nationwide cohort is relevant to people with a wide range of backgrounds and origins.

In the present study, the prevalences of SNHL and CHL and their hearing severity distributions across GI categories did not differ during the study period. In contrast, the WHO estimates a 5-fold increase in HL during the next three decades [2]. Other studies reported better hearing trends over this period, for example, over the 20 years in a nationwide Norway cohort study [38] and the study that showed a declining prevalence of HL in US adults aged 20 to 69 [39].

Hearing impairment was more significant among males than females, with ORs of 1.6 and 1.3 among those with SNHL and CHL, respectively. This is in accordance with WHO global data from 2018, which showed that males comprised 56% of those with hearing impairment [2].

We found that hearing impairment was highly associated with low GI among adolescents, thus rendering hearing impairment a potential risk factor for poor linguistic and communication-based skills [31]. A meta-analysis conducted by Purcell et al. [40] concluded that children with unilateral HL had lower full-scale and performance IQ scores than children with normal hearing. Second, GI is associated with education, SES, and parental literacy. The negative relation between HL in children and parental literacy [13] suggests that the relation observed between HL and GI may reflect the already recognized association between education and SES and the subsequent risk for HL [41]. However, after adjusting for these factors, the persistence of the relation between HL and GI suggests that the correlation is steady and statistically significant. Third, individuals with CHL are prone to educational gaps and lost school days due to chronic otitis media [42]. Moreover, due to the risk of tympanic perforation, surgery is not recommended for children with CHL before early school age, usually 9–13 years [43]. Thus, children may experience critical first years of school with impaired hearing. In addition, implementing national newborn hearing screening programs has improved the early management of children affected with SNHL by providing rehabilitative speech and language development [44].

Several studies have investigated the association of CHL with SES [44]. For example, among 6,424 preschool children from South African communities, representative of low- and middle-income countries, CHL was more common than SNHL or mixed HL in both females and males [42]. In addition, in our study, while SNHL showed a correlation between HL severity and GI, CHL did not. This might be attributed to the differences in management between the conditions, whereby SNHL management is severity-based, and CHL management is based on the cause, with less attention to hearing rehabilitation.

Our study has some inevitable limitations. First, the analysis was conducted primarily on adolescents within a specific age group, limiting the extrapolation of the results to young adults and the pediatric population. Second, only a limited number of variables associated with HL and GI (confounders) were considered and excluded, including psychiatric and neurologic comorbidities. Other potential confounders that were not considered include lifestyle and environmental factors (e.g., mobile phone use and recurrent acoustic trauma) and various comorbidities that might be related to HL and GI. Third, pure tone audiometry was not performed for all potential army recruits but was based on family physicians’ reports. Thus, an underestimation of hearing impairment might account for the relatively low prevalence of hearing impairment. Fourth, while our study provides insights into the relationship between HL and GI, a notable limitation is the absence of individual subtest scores within the GI test. This constraint precludes a more detailed exploration of how different aspects of cognitive function, particularly those unrelated to hearing and language, are specifically affected by different types of HL. Finally, several populations in Israel are underrepresented in this study due to military service exemption. These include females who are married, pregnant, mothers, or members of a minority; religious orthodox Jewish populations; and males of certain minorities (the Druze and Circassian communities are the only minorities that obligate their males to serve in the military, whereas males of other minorities can volunteer).

HL was correlated with decreased intelligence, as measured by GI. Furthermore, GI was less significantly impaired in individuals with SNHL than CHL. We suggest that tailored treatment should be offered for HL, based on cause and severity, with specific emphasis on early treatment and giving full attention to supportive linguistic, learning, and communicative functions.

The work has been approved by the IDF Medical Corps Institutional Review Board (protocol No. 2072-2020), which waived the requirement for written informed consent instead of its retrospective study design.

The authors have no conflicts of interest to declare.

No funding was received for this study.

N.T.: project administration, conceptualization, data curation, methodology, and writing – original draft.

Y.R.: methodology, formal analysis, and writing – original draft.

M.R.: visualization, investigation, and writing – original draft.

S.A.: writing – original draft, formal analysis, and validation.

A.L.: data curation, formal analysis, and writing – review and editing.

O.H.: writing – review and editing and conceptualization.

Z.B.: supervision, conceptualization, data curation, methodology, and writing – review and editing.

Nir Tsur and Yonatan Reuven contributed equally to the study.

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.