Association of serum 25-hydroxyvitamins D2 and D3 with hearing loss in US adults: analysis from National Health and Nutrition Examination Survey, 2015–2016

Background Hearing loss (HL) is increasingly recognized as a significant global public health issue, and research on its relationship with vitamin D levels has gained wider attention. However, the association between serum biomarkers 25-hydroxyvitamin D2 (25(OH)D2) and D3 (25(OH)D3) with different types of HL remains unclear. This study aimed to investigate the potential association of serum 25(OH)D2 and 25(OH)D3 with HL in US adults. Methods A sample of 3,684 individuals aged 20–69 years from the 2015–2016 National Health and Nutrition Examination (NHANES) was analyzed in this study. HL was defined as a pure tone average > 25 dB in either ear at low frequencies (500, 1,000, 2000 Hz), speech frequencies (500, 1,000, 2000, 4,000 Hz), and high frequencies (3,000, 4,000, 6,000, 8,000 Hz). Logistic regression was employed to examine the association between serum 25(OH)D2 and 25(OH)D3 and HL. The study population was then stratified by age, gender, race, and education level to analyze potential differences between adults in different subgroups. Results In the multivariate analysis, it was found that serum 25(OH)D2 was independently associated with low-frequency hearing loss (LFHL) (OR: 1.012 [95% CI, 1.005–1.020]) and speech-frequency hearing loss (SFHL) (OR: 1.011 [95% CI, 1.003–1.018]). Restrictive cubic spline analysis demonstrated a linear dose–response relationship between serum 25(OH)D2 levels and LFHL (p for linearity <0.001), as well as SFHL (p for linearity = 0.001). Conversely, an L-shaped association was observed between serum 25(OH)D3 levels and both LFHL (p for nonlinearity = 0.014) and SFHL (p for nonlinearity = 0.025), with threshold values identified at 35.3 and 36.5 nmol/L, respectively. Higher levels of serum 25(OH)D3 were associated with a lower probability of high-frequency hearing loss (HFHL) (OR: 0.994 [95% CI, 0.989–0.999]), with a threshold value identified at 53.9 nmol/L. Furthermore, a significant interaction between diabetes and serum 25(OH)D2 in LFHL was revealed through subgroup analysis (p = 0.041). In the non-diabetic population, serum 25(OH)D2 maintained its association with LFHL. Conclusion Our findings suggested a positive association between serum 25(OH)D2 concentrations and both LFHL and SFHL in the studied cohort. Additionally, an L-shaped relationship was found between serum 25(OH)D3 and LFHL and SFHL, and higher levels of serum 25(OH)D3 were identified to be associated with a lower risk of HFHL.


Introduction
Hearing loss (HL) is increasingly recognized as an important public health issue in the contemporary age, affecting more than 5% of the world's population (1).Multiple reasons have been attributed to the escalating prevalence of HL, including an aging population, noise from occupational or recreational settings, and the widespread use of headphone devices (2).HL not only hinders interpersonal interactions, but also exerts a profound impact on personal quality of life, and daily functioning (3)(4)(5), posing enormous health and economic burdens.Therefore, it is of great necessity to identify potential risk factors for potential HL.
The role of vitamin D in auditory impairment has increasingly become a focal point of research.Studies have revealed (6) that a deficiency in vitamin D can lead to the demineralization of cochlear calcium and disrupt microcirculation, which in turn contributes to changes in cochlear morphology and the incidence of HL.Further experimental investigations (7) have confirmed that vitamin D plays a critical role in the proliferation and differentiation of neural stem and progenitor cells, which is essential for preserving the normal function of the auditory nerve.In animal experiments, zebrafish embryos deficient in vitamin D receptor (VDR) gene were found to produce fewer sensory hair cells in their ears, resulting in HL and motor balance disorders (8).Similarly, mice lacking the VDR gene exhibited hypomineralization in their auditory ossicles, impairing sound transmission through the middle ear (9).Recent research has also demonstrated that VDR gene regulates brain natriuretic peptide via the cGMP-PKG signaling pathway, promoting neurite outgrowth and survival of cochlear spiral ganglion neurons (10).This growing body of evidence underscores the significance of vitamin D in maintaining auditory health.
In epidemiological investigations, to date, only a handful of studies have linked vitamin D with HL (11)(12)(13)(14).A cross-sectional study revealed a relationship between vitamin D deficiency and an increased likelihood of low-frequency hearing loss (LFHL) and speech-frequency hearing loss (SFHL) among individuals aged 70 and over (13).Another study, based on data from the UK Biobank, found that vitamin D intake is negatively correlated with the incidence of HL (15).Several other research concentrated on the relationship between vitamin D and HL in specific populations; for example, vitamin D deficiency has been deemed as a risk factor for HL in patients with diabetes (11).However, more recent studies indicate that diabetes is correlated with high-frequency hearing loss (HFHL) (16).Therefore, since vitamin D deficiency is prevalent in diabetic patients (17,18), HL may not be directly caused by diabetes but could be due to vitamin D deficiency in type 2 diabetes (T2DM) patients.This indicates that the current evidence is not certain and requires further investigation.
Given the backdrop of prior studies, it appears apparent that research revolving around the correlations between vitamin D and HL has primarily focused on the elderly or specific populations such as diabetics.Furthermore, the majority of studies have only focused on serum 25 Therefore, the current study aims to clarify the association between serum 25(OH)D 2 and 25(OH)D 3 and HL separately, as well as to identify their optimal concentrations for potential clinical reference values.

Study participants
National Health and Nutrition Examination Survey (NHANES) (20) is a national, cross-sectional survey that collects health-related data from the US population every 2 years.The program is well representative undergoing a 4-stage rigid design and includes data derived from in-depth interviews, series of physical examinations, and laboratory test results.Detailed information on the study design, methodology, and data collection can be found on the official NHANES website.
The current study utilized data of participants aged 20-69 years in NHANES from 2015 to 2016.Individuals were excluded if (1) audiometry data (N = 504) or abnormal tympanogram results, as well as those with type B and C tympanogram results were unavailable (21,22) (N = 361); (2) data of other covariates such as body mass index (BMI), hypertension, diabetes, history of stroke, history of congestive heart failure, smoking, firearms noise exposure, occupational noise exposure, recreational noise exposure, and supplement use were missing (N = 44); (3) without serum vitamin D level data (N = 174).A total of 3,684 eligible participants remained for final analysis.The participant screening flowchart is presented in Figure 1.

Vitamin D levels
The NHANES 2015-2016 cycle used a HPLC-MS/MS method to measure serum 25(OH)D 2 and serum 25(OH)D 3 levels.When the 25(OH)D 2 concentration is below the lower limit of detection, NHANES proposes using the lowest 25(OH)D 2 concentration (1.45 nmol/L) as the input value (23,24).For specific operations and details, refer to the NHANES Assessment of Vitamin D Levels document.

Audiometry
The audiometric requirements were met by all adults aged 20-69 years (25).The audiometric test was conducted by trained examiners in a soundproof room at the Mobile Examination Center (MEC) using an Interacoustics model AD226 audiometer with standard TDH-49P headphones and Etymotic EarTone 3A plug-in headphones to measure pure-tone air-conduction audiometry in both ears.Middle ear testing was performed using the Interacoustics Titan.Detailed instructions for this procedure can be found on the website (26).
HL was categorized as low-frequency, speech frequency, or high-frequency: LFHL defined as a mean of pure tones greater than 25 dB measured in either ear at 500, 1000, and 2000 Hz; SFHL defined as a mean of pure tones greater than 25 dB measured in either ear at 500, 1000, 2000, and 4,000 Hz; HFHL defined as a mean of pure tones greater than 25 dB measured in either ear at 3000, 4000, 6,000, and 8,000 Hz (13).

Covariates
Building upon previous research (25,27,28), we included the following covariates in our analysis: age, sex, race, level of education, BMI, hypertension, diabetes, history of stroke, history of congestive heart failure, smoking status, firearm noise, occupational noise, recreational noise, and supplement use.Definitions for medical histories including hypertension, diabetes, stroke, congestive heart failure, smoking status, and supplement use were all acquired through self-reports.Considering the potential impact of noise exposure on hearing, our analysis utilized three different sources of noise exposure, with definitions aligning with those found in prior literature (29).

Statistical analysis
Continuous variables were described as mean ± standard deviation or median (interquartile range) after Kolmogorov-Smirnov test for normality, and categorical variables as percentages.Depending on the characteristics of the variables, group differences were compared using the T-test, non-parametric tests, or chi-square tests.Multivariable logistic regression was employed to assess the relationship between serum 25(OH)D 2 , D 3 , and HL, followed by Generalized additive models (GAM) to further examine the nonlinear relationship.GAM provide flexibility by not assuming a predefined form for the relationship between the predictor and outcome variables.If a nonlinear relationship was detected, we applied piecewise logistic regression to identify threshold effects.This method involves dividing the predictor variable into segments and fitting separate logistic regression models within each segment.Restricted cubic splines allow for flexible modeling of nonlinear relationships by using piecewise polynomial functions that are smooth at the points where they join (knots).We visualized the relationship between serum 25(OH)D 2 and HL using restricted cubic splines with three knots; and for the relationship between serum 25(OH)D 3 and HL, five knots were used to further demonstrate an L-shaped relationship.Stratified logistic regression models were used for subgroup analyses, and likelihood ratio tests were applied to assess the effects and interactions within subgroups.All statistical analyses were performed using R version 4.3.0(30) and EmpowerStats (31) software.A two-sided p-value <0.05 was considered statistically significant.

Baseline characteristics of population with hearing loss
Among 3,684 participants aged 20-69 years, 328 had LFHL, 496 had SFHL, and 1,276 had HFHL.Those with older age, gender of male, Mexican-American ethnicity, lower education level, higher BMI, hypertension, diabetes, history of congestive heart failure, history of stroke, smoking, supplement use, occupational and recreational noise exposure were more likely to display HL in all three types (p < 0.05).Firearm noise exposure was found to be only associated with SFHL and HFHL (p < 0.05).
Across three types of HL, vitamin D levels showed different characteristics.Those with serum 25(OH)D 2 concentration greater than 1.45 nmol/L were more inclined to be LFHL (p < 0.05), whereas serum 25(OH)D 3 concentration was higher in SFHL and HFHL group (both p < 0.05) (Table 1).

Threshold effects and restricted cubic splines
To further explore whether a nonlinear relationship existed between serum 25(OH)D 2 , D 3 , and HL, we implemented piecewise logistic regression using GAM and recursive methods to calculate effect thresholds.Restricted cubic splines were then used to visualize the relationship.We found a linear dose-response relationship between serum 25(OH)D 2 and LFHL (p for linearity <0.001) and SFHL (p for linearity = 0.001).Comparatively, serum 25(OH)D 3 had an L-shaped relationship with LFHL (p for nonlinearity = 0.014) and SFHL (p for nonlinearity = 0.025), with critical values of 35.3 and 36.5 nmol/L, respectively (Figure 2 and Table 3).
When it comes to the part of serum 25(OH)D 3 , for Mexican-Americans and those with a high school education level, it was found to be associated with smaller likelihood of SFHL (p < 0.05).Moreover, in females and those with a BMI 25-30 subgroup, 25(OH)D 3 was associated with smaller likelihood of HFHL (p < 0.05).No association

Discussion
Utilizing a considerable sample size from the NHANES 2015-2016 cycle, we investigated potential relationships between serum 25(OH)D 2 , D 3 levels and HL in adults.Major findings suggested that serum 25(OH) D 2 had a positive independent association with both LFHL and SFHL.Restricted cubic splines confirmed a linear dose-response relationship between serum 25(OH)D 2 and these forms of HL.Interestingly, an L-shaped relationship was observed between serum 25(OH)D 3 and both LFHL and SFHL, with critical values of 35.3 nmol/L and 36.5 nmol/L, respectively.Higher levels of serum 25(OH)D 3 were associated with a lower likelihood of HFHL, with a critical value of 53.9 nmol/L.Subgroup analysis revealed a significant interaction between diabetes and serum 25(OH)D 2 in LFHL outcomes, maintaining its significant association within the non-diabetic population.
Previous studies have primarily focused on the association between serum 25(OH)D concentration and HL (11,12,32,33).One study (13) using a sample of 1,123 older adults from NHANES (2005-2010) suggested that low serum 25(OH)D concentration was positively correlated with the occurrence of LFHL and SFHL but not HFHL.This cross-sectional study's findings align with our observation of an L-shaped relationship between serum 25(OH)D 3 concentration and HL.Another NHANES study (14) also tried to establish a correlation between vitamin D deficiency and bilateral low-frequency hearing impairment and sensorineural HL.Other studies have revealed a negative correlation between serum 25(OH)D 3 and depression, while serum 25(OH)D 2 has an inverted U-shaped relationship with depression (34).Additionally, serum 25(OH)D and 25(OH)D3 concentrations are positively correlated with cognitive function (23), all linked to mental disorders.
The L-shaped relationship between serum 25(OH)D 3 and HL could be explained by several mechanisms.To begin with, the presence of vitamin D deficiency might contribute to HL by affecting cochlear bone density (35), leading to cochlear demineralization and microcirculation changes, altering calcium metabolism, and causing morphological changes in the cochlea that eventually result in hearing decline (6).Another underlying mechanism could be that vitamin D can cross the blood-brain barrier (36) and distribute unevenly among various brain regions (37).This uneven distribution affects the proliferation and differentiation of neural stem cells and progenitors (7), leading to impaired neuronal proliferation and differentiation, affecting vestibuloauditory neural function and resulting in hearing decline if vitamin D is deficient.
The positive independent association identified between serum 25(OH)D 2 and low and SFHL can be putatively explained by the interaction between vitamin D2, D3.Previous research (38,39) has shown that vitamin D2 supplementation can decrease circulating concentrations of 25(OH)D 3 , potentially as a result of affecting the metabolism of vitamin D3.This might be partly explained by the competition between vitamin D2 and D3 for the same hydroxylation pathways, with vitamin D3 being a more effective factor in maintaining and elevating serum total 25(OH)D levels (40).In light of this, high levels of vitamin D2 might competitively affect the activation of Subgroup analysis observed a significant interaction between diabetes and serum 25(OH)D 2 in LFHL.Previous research (41) has shown that vitamin D are negatively associated with the risk of T2DM, potentially because vitamin D plays a underlying part in regulating the process of immune and inflammatory cells proliferation, differentiation and function, upregulating anti-inflammatory pathways and downregulating the activation of these cells, thereby reducing the risk of T2DM (42).Longitudinal studies have also correlated diabetes with an increased risk of moderate or severe HL (43).
This study has several strengths.Firstly, we separately investigated the associations between serum 25(OH)D 2 and 25(OH)D 3 , and adult HL, uncovering contrasting trends between these markers.To our knowledge, this study is the first to explore the association in this perspective, as previous research was limited to seeking the association between total serum 25(OH)D and HL.Our research underscores the necessity of individually considering serum 25(OH)D 2 and 25(OH)D 3 when studying the correlation between vitamin D and HL.Secondly, the large sample size provided by NHANES ensured the reliability of our statistical results.Restricted cubic splines and threshold analysis assessed the L-shaped relationship between serum 25(OH)D 3 and LFHL and SFHL.The primary limitation of this study lies in its crosssectional design, which precludes the determination of causality.Additionally, due to database constraints, it was not possible to adjust for all potential confounding variables, such as osteoporosis and parathyroid hormone levels, which may influence vitamin D levels and auditory health.Moreover, when the 25(OH)D 2 concentration is below the detection limit, we use the minimum 25(OH)D 2 concentration (1.45 nmol/L) as the input value.This does not reflect the patient's true 25(OH)D 2 concentration.Assessing vitamin D levels based on a single test may be insufficient to fully capture its impact on HL, and further research is needed to determine the precise effects and implications of 25(OH)D 2 at accurate concentrations.Based on our findings, which provide potential epidemiological evidence for differing vitamin D supplementation categories, further prospective research relating to this topic is encouraged to establish causality between these factors.

Conclusion
This study consolidated that in the adult population of the United States, serum 25(OH)D 2 concentration was positively associated with LFHL and SFHL, while serum 25(OH)D 3 exhibited an L-shaped relationship with LFHL and SFHL, with critical values of 35.3 and 36.5 nmol/L, respectively.Higher levels of serum 25(OH)D 3 were associated with a lower likelihood of HFHL, with a critical value of 53.9 nmol/L.Furthermore, serum 25(OH)D 2 retained its association with LFHL in the non-diabetic population.
Future research should focus on longitudinal and interventional studies to establish causality and explore the effects of vitamin D supplementation on HL.Mechanistic studies are needed to understand the biological processes involved, while studies on diverse populations can help determine the universal applicability of the findings.Comparative studies between vitamin D2 and D3, as well as genetic studies to explore individual responses, are also recommended.
-hydroxyvitamin D (25(OH)D) or dietary vitamin D intake concerning HL, yet the individual contributions of serum 25-hydroxyvitamin D 3 (25(OH)D 3 ) and D 2 (25(OH)D 2 ) have been less studied.Since serum 25(OH)D is composed of both 25(OH)D 3 and 25(OH)D 2 , it is important to distinguish between the two.Vitamin D 3 (cholecalciferol) is primarily synthesized in the skin through sunlight exposure, whereas vitamin D 2 (ergocalciferol) is mainly obtained from plant-based foods and supplements.Both forms undergo hydroxylation in the liver to form 25(OH)D 3 and 25(OH)D 2 , respectively, which are the main circulating forms.Currently, serum 25(OH)D or 25(OH)D 3 are the primary methods for assessing vitamin D status.Measuring 25(OH)D 2 can provide a more comprehensive assessment of vitamin D status (19).With the use of high-performance liquid chromatography-tandem mass spectrometry (HPLC-MS/MS), 25(OH)D 2 has further refined the classification of these measurements.

FIGURE 1
FIGURE 1The flowchart of the participants' selection process.

TABLE 1
Characteristics of participants classified by hearing loss.

TABLE 2
Multivariate logistic regression models assessing the relationship between 25-hydroxyvitamin D 2 , D 3 and hearing loss.

TABLE 3
The results of two-piecewise logistic regression model.hindering the body's ability to maintain adequate levels of total serum 25(OH)D, and ultimately lead to the occurrence of HL.