Conicity Index Associated with Adult Hearing Impairment: A Cross-Sectional Study Based on the National Health and Nutrition Examination Survey (2001–2018)

Abstract

Objective:

This study aimed to evaluate the association of conicity index (C-index), a marker of abdominal obesity, with hearing impairment (HI) in US adults using data from the National Health and Nutrition Examination Survey (NHANES).

Methods:

We conducted a cross-sectional investigation using data from nine NHANES cycles spanning from 2001 to 2018. The C-index was used as the independent variable, whereas HI—defined as a pure-tone average hearing threshold exceeding 25 dB HL—was the outcome measure. Multivariable logistic regression models were used to assess the association between C-index and HI, adjusting for potential confounders including age, sex, race/ethnicity, hypertension, diabetes, total cholesterol, haemoglobin and blood urea nitrogen. Dose–response relationships were evaluated using trend tests (P for trend) and restricted cubic spline (RCS) analysis. Subgroup analyses were performed to assess potential interaction effects.

Results:

A total of 7394 adults aged 20–69 were included. Higher C-index values were significantly associated with an increased prevalence of HI. In the fully adjusted model, each unit increase in C-index was associated with 7.35-fold higher odds of HI (95% CI: 3.03–17.82, P < 0.01). When stratified into tertiles, individuals in the highest tertile had a 57% higher likelihood of HI compared to the lowest tertile (Odds Ratio = 1.57, 95% Confidence Interval: 1.14–2.18, P < 0.01). RCS analysis indicated a significant non-linear association (P for non-linearity < 0.05) with an inflection point at a C-index of 1.17. Subgroup analysis revealed significant interaction effects of C-index with gender, diabetes and hypertension (P for interaction < 0.05) but not with age (P > 0.05).

Conclusion:

Higher C-index values are independently associated with greater odds of HI. Given the rising burden of obesity and its metabolic consequences, the C-index may serve as a practical tool for HI risk assessment, especially in populations with cardiometabolic risk factors.

KEY MESSAGES

• (1)The conicity index (C-index) is a significant predictor of hearing impairment (HI), with higher values indicating increased odds of HI.
• (2)A non-linear association exists between the C-index and HI, with a risk inflection point at 1.17.
• (3)Subgroup analyses reveal significant interactions with gender, hypertension, and diabetes.
• (4)Monitoring the C-index could guide interventions to prevent HI in at-risk populations.

INTRODUCTION

Hearing impairment (HI) affects approximately 1.57 billion people worldwide and ranks as the fourth leading cause of disability.[1,2] Despite its impact, routine screening for HI remains infrequent due to high costs and a common perception of low urgency among affected individuals, leading to delayed diagnosis and treatment.[2,3]

Obesity, another major public health issue, has been strongly linked to metabolic and cardiovascular conditions, such as diabetes and hypertension, which have been identified as potential risk factors for HI.[4,5,6,7,8] Biological mechanisms underlying this association are well-supported; that is, obesity-related metabolic disorders may lead to reduced microvascular circulation, diminishing cochlear blood flow, damaging hair cells and ultimately leading to hypoxic injury.[9,10,11]

Fat distribution, rather than overall obesity, appears to play a critical role in metabolic health.[12,13,14] The conicity index (C-index) is an emerging anthropometric measure of abdominal obesity and offers a more refined assessment than body mass index, waist circumference (WC), or waist-to-hip ratio.[15] Higher C-index values are strongly associated with vascular dysfunction and cardiometabolic risks, which may impact hearing.[16,17,18] However, the association between C-index and HI remains unexplored.

To address this research gap, we conducted a cross-sectional analysis using nine National Health and Nutrition Examination Survey (NHANES) cohorts to investigate the C-index–HI association while controlling for key covariates related to vascular health, metabolism and auditory function.[19,20,21,22] Understanding this association may provide insights into preventive strategies for HI, particularly among individuals at risk for metabolic disorders.

MATERIALS AND METHODS

Study design and participant selection

This cross-sectional study used publicly available data from NHANES (https://wwwn.cdc.gov/nchs/nhanes/), which was conducted by the U.S. National Centre of Health Statistics (NCHS) and Centers for Disease Control and Prevention. The NHANES program employs a complex multi-stage probability sampling design to generate a dataset representative of the non-institutionalised US population. As NHANES data are publicly available and de-identified, additional ethical approval was not required.

Data from nine NHANES cycles (2001–2018) were included, and these cycles provided comprehensive audiometric and anthropometric data required for analysis. The inclusion criteria were as follows: adults aged 20–69 years (to focus on non-elderly adult HI); participants with complete audiometry data; availability of WC, weight, and height measurements for C-index calculation; and complete data for key covariates (age, gender, race, hypertension, diabetes, blood cholesterol, haemoglobin and urea nitrogen). The following strict exclusion criteria were applied: participants with missing or unreliable audiometric assessments; individuals with self-reported ear infections, exposure to ototoxic drugs, or a history of ear surgery; and participants with occupational noise exposure history (as noise-induced hearing loss could confound results). After implementing these criteria, 7394 eligible participants from an initial pool of 91,351 were included in the final analysis [[Figure 1].

Figure 1.

Flowchart of data enrolment in this study.

C-index measurement

The C-index, used as the exposure variable in this study, was calculated from WC, weight, and height measurements,[18] following the formula: 0.109⁻¹ × WC (m) × [weight (kg)/height (m)]^−1/2.[13] All anthropometric measurements were conducted by certified health technicians at a Mobile Examination Centre using standardised protocols.

Definition of HI

HI was defined as an average pure-tone hearing threshold exceeding 25 dB HL in the better-hearing ear, consistent with previous epidemiological studies.[23,24] Audiometric testing was conducted in a soundproof mobile unit by using a calibrated audiometer to measure air conduction thresholds at 0.5, 1, 2, 3, 4, 6 and 8 kHz. Instrumentation for the audiometry component included an Audiometric Research Tool (ART) system with standard TDH-49P headphones (Telephonics Corporation, CA, USA) and Etymotic EarTone 3A insert earphones (Lucid Hearing Holding Company, TX, USA). Data from the ART system were captured electronically and uploaded into the NHANES database automatically. To ensure accuracy, each ear was tested twice at 1 kHz to assess test reliability; if discrepancies >10 dB HL were observed, then additional verification was performed.[25,26]

Selection of covariates

A multivariate-adjusted model was developed to account for potential covariates. The following covariates were included in the analysis based on their biological relevance to both obesity and HI: (1) demographic variables: age, gender and race; (2) age: <44 years or ≥44 years; (3) race: Non-Hispanic White and Non-Hispanic Black or others and (4) health conditions: (i) hypertension and diabetes: based on the medical history filled in by the respondents themselves; (ii) blood cholesterol: measured using an enzymatic assay (normal range: <200 mg/dL); (iii) haemoglobin: analysed via a Beckman Coulter DxH 800 instrument (Beckman Coulter, Inc., USA), normal: 13.8–17.2 g/dL for men; 12.1–15.1 g/dL for women and (iv) urea nitrogen: quantified using high-performance liquid chromatography (normal: 2.5–7.1 mmol/L). The measured values for these biomarkers are cholesterol 77–545 mg/dL, haemoglobin 6.4–19.9 g/dL and urea nitrogen 0.71–28.2 mmol/L.

Statistical analysis

Continuous variables were first tested for normality using the Shapiro–Wilk test. Variables following a normal distribution were expressed as mean ± standard deviation (x̄ ± s), while non-normally distributed variables were presented as median and interquartile range (P25, P75). Categorical variables were expressed as frequencies and percentages.

Group comparisons for continuous variables were conducted using independent sample t-test for normally distributed data and Mann–Whitney U test for non-normally distributed data. For categorical variables, chi-square test was used.

Adjustments were made to account for clustering and stratification effects because of the complex multiple-stage probability sampling strategy design of the NHANES. To assess the robustness and predictive ability of our logistic regression models, we conducted model fitting evaluations: (1) Hosmer–Lemeshow test: used to evaluate the goodness-of-fit of the logistic regression models (P > 0.05 indicates a well-fitting model); (2) Akaike Information Criterion (AIC): a method of assessing how well a model fits the data it generates, with lower values indicating a better fit; (3) area under the Receiver Operating Characteristic curve (AUC): measures discrimination ability (values closer to 1.0 indicate stronger predictive power). The C-index values were categorised into tertiles, and binary logistic regression models were employed to evaluate the associations of the C-index with HI. Three progressively adjusted models were created for comprehensive analysis. Model 1 was left unadjusted. Model 2 was adjusted for demographic variables (age, gender, and race). Model 3 was adjusted for all covariates, including age, gender, race, diabetes, hypertension, blood cholesterol, haemoglobin and urea nitrogen. Analyses of subgroups defined by age, gender, hypertension and diabetes were conducted using the fully adjusted model. Interaction tests were performed to evaluate the degree of heterogeneity within these subgroups. Data analysis and drawing were conducted by using R 4.3.2 (University of Auckland, New Zealand) and DecisionLinnc1.0 software (Hangzhou Apes Information Technology Co., Ltd., China). A visual interface was used to facilitate data processing, analysis, and machine learning in DecisionLinnc1.0, a platform that unifies various programming language environments.[27] A two-tailed P-value < 0.05 was considered statistically significant.

RESULTS

Baseline characteristics of the study cohort

Among the 7394 individuals included, the mean age was 44.00 (31.00–56.00) years, with 20.85% overall prevalence of HI. The gender distribution was balanced, comprising 49.66% females and 50.34% males. Participants aged over 44 years accounted for 58.89% of the cohort, while those under 44 comprised 41.11%. Regarding racial distribution, 32.73% of participants were non-Hispanic whites, 23.72% were non-Hispanic blacks, and 43.55% belonged to other racial groups. Statistically significant differences (P < 0.01) were observed across gender, race, age, hypertension, diabetes, blood cholesterol, haemoglobin, urea nitrogen and C-index when stratified by hearing status. Males had a higher incidence of HI compared to females. The HI group was notably older and demonstrated higher rates of diabetes and hypertension. Additionally, this group exhibited elevated levels of blood cholesterol, haemoglobin, urea nitrogen and C-index values compared with those without HI. Details can be found in the supplementary material Table S1.

Association of the C-index with HI

Table 1 demonstrates a positive association of the C-index with HI in the fully adjusted model (odds ratio [OR] = 7.35, 95% confidence interval [CI]: 3.03–17.82), indicating that for each unit increasing in the C-index, the likelihood of experiencing HI increased by 7.35 times. According to the model fitting evaluation, a satisfactory fit was shown by the Hosmer–Lemeshow test (P = 0.399). The AIC for a fully adjusted model is 1245.32, which is lower than unadjusted models, indicating improved performance. The AUC is 0.82, indicating that the model demonstrates excellent discriminatory power in identifying individuals with HI.

Table 1.

Association of the C-index with HI

	Model 1		Model 2		Model 3
C-index	OR (95% CI)	P-value	OR (95% CI)	P-value	OR (95% CI)	P-value
Continuous	933.97 (476.28, 1845.76)	<0.001	9.94 (4.24, 23.33)	<0.001	7.35 (3.03, 17.82)	<0.001
Tertile
Tertile 1	Reference	Reference	Reference	Reference	Reference	Reference
Tertile 2	3.99 (3.19, 5.05)	<0.001	1.24 (0.96, 1.61)	0.104	1.20 (0.93, 1.57)	0.161
Tertile 3	7.71 (5.90, 10.16)	<0.001	1.74 (1.27, 2.39)	0.001	1.57 (1.14, 2.18)	0.006
P for trend		<0.001		<0.001		<0.001

C-index, conicity index; Model 1: no adjustment; Model 2: adjusted for age, gender, and race; Model 3: adjusted for all potential confounders.

Notes: OR, odds ratio; CI, confidence interval.

The C-index was analysed as a three-level categorical variable derived from a continuous variable. The results showed that individuals in the highest tertile had a higher incidence of HI compared to the lowest tertile (OR = 1.57, 95% CI: 1.14–2.18, P for trend <0.01). This trend reinforces the observation that elevated C-index levels correspond to greater odds of HI.

Non-linear correlation between C-index and HI

Restricted cubic spline (RCS) analysis was used to illustrate the correlation between the C-index and HI [[Figure 2]. The results revealed a significant non-linear positive correlation between the C-index and HI. A threshold effect analysis and identified an inflection point at a C-index of 1.17 in Model 3. The analysis showed that when the C-index exceeded 1.17, the risk of HI increased sharply.

Figure 2.

The correlation between HI and C-index.

Note: Restricted cubic spline showed a non-linear correlation between the C-index and HI. Solid yellow represented the fitted regression line; 95% CI was shown by the blue region; OR equal to 1 was indicated by a grey dashed line. C-index, conicity index; HI, hearing impairment.

Subgroup analyses

Subgroup analysis was conducted to investigate the complex correlation between C-index and HI across different categories of gender, age, diabetes and hypertension [Table 2]. The results indicated statistically significant interaction effects between the C-index and HI with respect to gender, diabetes and hypertension (P < 0.05); no significant interaction was found for age (P > 0.05). These findings highlight that the association of the C-index with HI varies across specific subgroups, particularly influenced by gender, the presence of diabetes, and hypertension status.

Table 2.

Subgroup analyses for the association of the C-index with HI

Variable	OR (95% CI)	P-value	P for interaction
Gender			<0.001
Male	4016.31 (1527.16,10562.62)	<0.001
Female	131.49 (46.36,373)	<0.001
Age (year)			0.092
<44	9.49 (1.11,81.43)	0.04
≥44	86.29 (37.84,196.79)	<0.001
Diabetes			0.012
Yes	35.47 (5.91,212.89)	<0.001
No	494.51 (226.5,1079.63)	<0.001
Hypertension			0.039
Yes	102.12 (32.72,318.76)	<0.001
No	467.94 (184.78,1185.04)	<0.01

DISCUSSION

Our findings demonstrate a strong, nonlinear correlation between the C-index and HI via RCS analysis. This suggests that abdominal obesity may be a more critical risk factor for HI than overall body fat.

To our knowledge, this is the first study to specifically examine the association of C-index with HI. Previous research has explored the broader association between obesity and HI. Kim et al.[28] discovered in the mouse study that diet-induced obesity and metabolic changes reduced auditory sensitivity in male mice. These results suggest that managing visceral fat distribution could potentially reduce the risk of HI.

The C-index has been proposed as an index of body fat distribution, with the value increases in the accumulation of abdominal fat in the body, had a higher diagnostic accuracy for obesity.[29] Higher C-index values reflect visceral fat accumulation, which promotes chronic inflammation, leading to vascular dysfunction and cochlear hypoxia.[30,31] These metabolic disturbances have been linked to microcirculatory impairment and cochlear hair cell damage, thereby increasing HI risk.[32,33] The inner ear, known for its high metabolic rate, is particularly vulnerable, and reduced cochlear blood perfusion due to endothelial dysfunction can exacerbate damage.[34,35,36] These mechanisms include diabetic neuropathy brought on by damage to the auditory system, reactive oxidative stress and cochlear damage brought on by microangiopathy.[13,37]

The association of C-index with HI was more pronounced in males, diabetics and hypertensive individuals, consistent with studies showing worse vascular outcomes in these groups. A retrospective study of 2602 individuals found that gender differences may stem from varying adiponectin levels, with females generally exhibiting higher concentrations.[38] Adiponectin, an adipokine secreted by adipose tissue, has anti-apoptotic, microcirculation-regulating and hearing-protective properties.[39,40] Furthermore, oestrogen has been reported to provide protective effects on the auditory system in females.[41] The lack of an age interaction may be due to NHANES sampling limitations (age cap of 69 years) and strong independent effects of age on HI.[42,43]

Our findings highlight the importance of abdominal obesity as a modifiable risk factor for HI, particularly among high-risk subgroups (men, diabetics and hypertensive individuals). Given the global rise in obesity, monitoring C-index as part of routine metabolic assessments could help identify high-risk individuals before a significant auditory decline occurs. Interventions targeting visceral fat reduction (e.g., diet, exercise and metabolic control) may help mitigate HI risk. Clinicians should consider taking C-index as one of routine physical examinations, particularly for patients with metabolic disorders.

This study has several strengths. First, it is based on NHANES data, a national representative sample collected using standardised procedures. Additionally, the study controlled for multiple covariates and included subgroup analyses to assess the robustness of the association of the C-index with HI across different demographic groups.

This study also has limitations. The cross-sectional design precludes establishing a definitive causal correlation between C-index and HI. Additionally, only four factors (age, gender, diabetes and hypertension) were selected for subgroup analysis. This selection was based on biological relevance and prior literature, which identified these factors as key effect modifiers in the correlation between obesity and HI. Although other variables (e.g., cholesterol, haemoglobin and race) were adjusted for in the model, they were not included in the subgroup analysis due to a lack of significant interaction effects in preliminary analyses. Future studies should explore additional subgroup analyses in larger populations. Lastly, despite adjusting for multiple covariates, unmeasured variables such as medication use, occupational noise exposure and dietary habits could still influence the observed associations.

CONCLUSION

High C-index values are significantly associated with increased risk of HI, with a nonlinear correlation and stronger effects in males, diabetics and hypertensive individuals. These findings underscore the importance of addressing abdominal obesity as a modifiable risk factor for HI. Early screening and intervention strategies may help reduce the burden of HI, particularly in at-risk populations.

Availability of data and materials

The National Health and Nutrition Examination Survey (NHANES) database hosts all study datasets, which are openly available through its official portal: https://wwwn.cdc.gov/nchs/nhanes/.

Author contributions

Guifen Cai: Research architecture formulation, study protocol orchestration, primary data procurement, analytical modeling and hermeneutics and initial manuscript composition. Xiaoyan Xu: Designed research framework and executed study protocols. Xiaokun Li: Verified outcomes and supervised project implementation. Qiong Fang: Conceptualized the study, revised the manuscript with critical insights. All authors critically reviewed the manuscript, approved the final version and assumed full responsibility for its integrity.

Ethics approval and consent to participate

The National Center for Health Statistics (NCHS) Research Ethics Review Board reviewed and approved all NHANES protocols involving human participants, with written informed consent obtained from each subject. As the current study constitutes a secondary analysis of NHANES data, it was granted an exemption by the institutional ethics committee.

Conflicts of Interest

The authors have no conflicts of interest to declare.

Funding Statement

None.