To investigate the effect of central obesity on the severity and characteristics of age-related hearing impairment (ARHI), we recruited 690 adult subjects with normal or symmetrical sensorineural hearing loss (SNHL). The effects of age, gender, morphometry, habits, systemic diseases, and environmental noise exposure on average pure tone hearing level at low frequencies (pure tone audiometry (PTA)-low) and high frequencies (PTA-high) were analyzed. After adjusting for age, gender, systemic disease, and other variables, waist circumference (WC) showed a significant positive association with PTA-low and PTA-high. In females, PTA-low and PTA-high only showed significant positive association with age, but not with WC or other variables. However, PTA-high showed a positive association with borderline significance with WC in female subjects older than 55. In males, WC as well as age and noise exposure showed significant positive associations with both PTA-low and PTA-high, primarily in subjects younger than 55. When both WC and BMI were taken into account in a backward stepwise multivariate linear regression analysis, WC, but not BMI, showed a significant positive association with PTA-low and PTA-high in males younger than 55, and with PTA-high with borderline significance in females older than 55. However, the audiogram patterns were not significantly affected by central obesity in either age or gender. Our results suggest that WC was, even after adjustment for BMI, an independent risk factor of ARHI, particularly for low and high frequencies in males younger than 55 and for high frequencies in female subjects older than 55.
Age-related hearing impairment (ARHI) is a complex disorder characterized by progressive deterioration of auditory sensitivity associated with aging. Contributing mechanisms may include hypoxia/ischemia, reactive species formation, oxidative stress, and apoptotic/necrotic death of hair cells and/or spiral ganglion cells. Inherited and acquired mutations in the mitochondrial DNA are also related to development of ARHI (1). In addition to genetic susceptibility, environmental factors, such as systemic diseases, noise, chemical exposure, tobacco, alcohol, ototoxic medication, diet, hormonal factors, and socioeconomic status, may also contribute to the variation in the frequency and severity of hearing loss across societies and among individuals; however, current data on some of these factors are contradictory (2).
On the other hand, obesity and its comorbidities including hypertension (HTN), type 2 diabetes mellitus, and dyslipidemia have emerged as a global epidemic (3). Prospective data have shown that obesity can increase metabolic risk, which subsequently increases the risk of not only coronary artery disease but also cerebrovascular disease (3). In addition, obesity was also associated with gastroesophageal reflux disease (4), pulmonary dysfunction in elderly people (5), gallbladder disease resulting in hospitalization (6), urinary incontinence in older women (7), early aging (8), Alzheimer's disease (9), and mortality (10).
The comorbidities of obesity have also been reported to be independently associated with ARHI (11,12,13). However, it is still unclear whether obesity itself is an independent risk factor for ARHI. Barrenäs et al. (14) reported that compared with subjects who were born small for their gestational age with subsequent catch-up growth, those born small without catch-up growth exhibited 134% higher risk of sensorineural hearing loss (SNHL) (14). Adult short stature was also associated with a 50% increased risk. Compared with subjects of average BMI, being overweight was associated with 30%, obesity with 99%, and overweight plus low birth weight for gestational age was with 118% higher risk of SNHL (14). In addition, among noise-exposed individuals with less than average height, increased age and HTN both had a negative impact on high frequency hearing thresholds. However, among tall employees, HTN had no effect on hearing and the influence of age was less pronounced (14). More recently, Fransen et al. (15) reported that taller people had better hearing on average, with a more pronounced benefit at low sound frequencies (<2 kHz), and that a high BMI correlated with hearing loss across all frequency ranges. Thus, this current evidence suggests that obesity may be closely associated with hearing function.
According to a consensus statement from the International Diabetes Federation, central obesity may be a more important risk factor than BMI for defining metabolic syndrome (3). However, the effects of central obesity on the severity of ARHI have never been reported. As for audiometric configurations, Schuknecht (16) first introduced three audiogram patterns (flat, down-sloping, and abrupt high-tone loss) subjectively for aged subjects and tried to link them to different kinds of age-related histopathologic changes of inner ear. The association between environmental risk factors and frequency characteristics (or audiometric patterns) of hearing loss was also reported. For example, exposure to firearm noise was associated with high frequency loss, people with diabetes and heavy smoker had significantly poorer hearing level at both low and high frequencies, and subjects with HTN had poorer hearing level at 1 kHz only (17). But, the characteristics or audiometric configurations of central obesity–related hearing impairment were never explored. Therefore, we try to address these issues in this study. We hypothesized that central obesity itself is an independent risk factor for ARHI and central obesity has its characteristic audiometric configuration.
Methods and Procedures
From January 2007 to August 2008, 762 adult subjects with normal to symmetric SNHL were recruited from the Health Management Department of National Taiwan University Hospital and from Dalin Tzu Chi General Hospital. All subjects received otoscopy, tympanogram, and a pure tone audiometric examination. Their medical history was obtained using a questionnaire used for annual health checkups in the Health Management Department of National Taiwan University Hospital. The exclusion criteria included age younger than 35, pregnancy, and asymmetric SNHL (defined as ≥15 dB HL asymmetry in ≥2 frequencies) (18). Subjects with any of the following histories were also excluded: external or middle ear disease, conductive hearing loss (presented with air-bone gap in audiogram), high environmental noise exposure and/or acoustic trauma (presented with 3–6 kHz dip in audiogram), exposure to an ototoxic drug, major neurological or psychiatric diseases, brain tumor or vestibular Schwannoma, vertigo (accompanied with nystagmus), liver cirrhosis, chronic renal failure treated with peritoneal dialysis or hemodialysis, cancer, head and neck radiation exposure, heavy smokers, alcoholism, or substance abuse. A total of 72 (9.4%) subjects were excluded from the 762 total subjects. Thus, 690 subjects were included in this analysis.
Height, body weight, and waist circumference (WC) were measured after an overnight fast. WC was measured in a horizontal plane, midway between the inferior margin of the ribs and the superior border of the iliac crest according to International Diabetes Federation guidelines (3). Central obesity in Asians is defined as a WC of >90 cm for males and >80 cm for females as described previously (3). The local hospital Institutional Review Board approved the study protocols and all subjects provided written informed consent.
Mean hearing level and audiometric configurations
Six frequencies were tested in a routine pure tone audiometric examination. First, the mean threshold of each frequency was calculated for each subject individually in both ears. Then, the mean individual subject thresholds of 250 Hz, 500 Hz, 1 kHz were averaged to obtain the average pure tone hearing level of low frequencies (PTA-low); and that of 2 kHz, 4 kHz, and 8 kHz as the average pure tone hearing level of high frequencies (PTA-high).
We classified the audiometric configurations of hearing impairment into four types according to the pattern: flat, downward sloping, abrupt high-tone loss, and reverse sloping. The former three types were classified subjectively by Schuknecht (16), and the last one is added subjectively by our study group. The rules for determining the various audiometric configurations were that “flat” type: violation of hearing threshold at all frequencies was within 10 dB HL; “downward sloping” type: average of mid-two frequencies (1 and 2 kHz) was ≥10 dB HL than that of lower-two frequencies (250 and 500 Hz) and smaller than higher-two frequencies (4 and 8 kHz); “abrupt high-tone loss “type: average of higher-two frequencies was ≥10 dB HL than that of mid-two frequencies, but the difference between the average of lower-two and mid-two frequencies was within 10 dB HL; “reverse sloping” type: average of lower-two frequencies was ≥10 dB HL than that of mid-two frequencies and that of mid-two frequencies was also larger than higher-two frequencies (19). All audiograms were fitted into one of these four patterns using force choice method by two authors, and the results were confirmed by the third author if agreement was not achieved initially.
The data are presented as the mean ± s.d., unless indicated otherwise. WC was treated as a dichotomous variable for audiogram pattern analysis (WC >80 cm for females; WC >90 cm for males). Student's t-test was used to test the difference of means between groups. Univariate linear regression and backward stepwise multivariate linear regression analysis were performed to test the impact of age, gender, WC or BMI, health-related habits (smoking and drinking), systemic diseases, and environmental noise exposure on PTA-low and PTA-high scores. Before age was entered into the linear equation, we confirmed that the relationship between PTA-low or PTA-high and age was linear but not quadratic in all age ranges, age <55, and age ≥55, in either sex.
We also used 2 × 4 Fisher exact test and χ2-test to test the distribution of audiogram patterns by age, gender, and central obesity. P < 0.05 was considered statistically significant. All analyses were performed using STATA 10.0 software (StataCorp, College Station, TX).
Of the 432 female and 258 male adult subjects, the mean age was 56.9 ± 8.5 years (range 35–79 years) for females and 57.1 ± 9.5 years (range 36–86 years) for males. The mean hearing level for low frequencies (PTA-low) was 16.5 ± 8.6 dB HL in females and 17.9 ± 11.6 dB HL in males. The mean hearing level for high frequencies (PTA-high) was 20.9 ± 13.0 dB HL in females and 32.6 ± 18.9 dB HL in males. The proportion of normal hearing (both PTA-low and PTA-high scores ≤ 25 dB hearing level) was higher in females (67.1%) than in males (42.2%).
The mean WC was 80.6 ± 8.5 cm in females and 87.3 ± 8.0 cm in males. Compared to the subjects without central obesity, the female-to-male ratio was higher in the central obesity group (2.38:1.26). Mean age, weight, and BMI were significantly higher in subjects with central obesity. In addition, HTN and diabetes mellitus were more prevalent in the central obesity group (Table 1). Regarding the hearing level, the PTA-low score was significantly higher in the central obesity group than in noncentral obesity group (17.9 ± 11.3 vs. 12.8 ± 5.8, P = 0.0029) in male subjects younger than 55. The PTA-high score was higher with borderline significance in the central obesity group of female subjects older than 55 (23.4 ± 12.4 vs. 26.6 ± 13.2, P = 0.0512) (Table 1 and Figure 1). These findings suggest that central obesity may be associated with hearing loss, but this association may differ by age and gender.
Table 1. The comparisons between nonobesity and central obesity subjects
Univariate linear regression demonstrated that age, WC, BMI, smoking, CAD, and chronic renal failure had significant positive associations with both PTA-low and PTA-high scores (Table 2). Male gender, height, weight, alcohol drinking, HTN, and noise showed significant positive associations with PTA-high, but not with PTA-low (Table 2). The positive association of WC and BMI with the PTA scores suggests that being overweight or obese may be associated with ARHI.
Table 2. Univariate linear regression analysis for PTA-low and PTA-high by all risk factors
Using a backward stepwise multivariate regression model adjusting for age, gender, systemic diseases, and the other variables, WC still showed a significant positive association with PTA-low (β ± s.e. = 0.09 ± 0.04, P = 0.034, adjusted R2 = 0.1561) and PTA-high (β ± s.e. = 0.14 ± 0.06, P = 0.024, adjusted R2 = 0.3589) (Table 3). In addition, we also noted that the association of WC with ARHI may differ by age and gender. In all females, PTA-low and PTA-high were only positively associated with age, but were not associated with WC and the other variables. However, in female subjects older than 55, WC was positively associated with PTA-high with borderline significance (β ± s.e. = 0.17 ± 0.09, P = 0.051, adjusted R2 = 0.2306) (data not shown). In all males, WC as well as age and noise exposure were positively associated with both PTA-low and PTA-high. In addition, alcohol drinking was also positively associated with PTA-high in males (Table 3). In contrast to females, WC was associated with both PTA-low (β ± s.e. = 0.47 ± 0.10, P < 0.001, adjusted R2 = 0.2433) and PTA-high (β ± s.e. = 0.70 ± 0.21, P = 0.001, adjusted R2 = 0.2006) in male subjects younger than 55, but not in males older than 55 (data not shown). In summary, WC was positively associated with both PTA-low and PTA-high scores in males younger than 55 and showed a borderline positive significance with the PTA-high score in females older than 55 in backward stepwise multivariate linear regression analysis.
Table 3. Backward stepwise multivariate linear analysis for PTA-low and PTA-high by waist circumference and other risk factors in all subjects and by genders
Using BMI instead of WC in backward stepwise multivariate linear regression model, BMI also showed a significant positive association with PTA-low (β ± s.e. = 0.23 ± 0.10, P = 0.019, adjusted R2 = 0.1587) (data not shown) but not with PTA-high, after adjusting for age, gender, systemic diseases, and other variables. In subgroup analyses, BMI only showed a significant positive association with PTA-low (β ± s.e. = 1.07 ± 0.26, P < 0.001, adjusted R2 = 0.2162) and PTA-high (β ± s.e. = 1.16 ± 0.55, P = 0.035, adjusted R2 = 0.1488) in males younger than 55 (data not shown). BMI was not significantly associated with either PTA-low or PTA-high in males older than 55 or in females younger or older than 55 by backward stepwise multivariate linear regression (data not shown).
Furthermore, when both WC and BMI together were taken into account in backward stepwise multivariate linear regression analysis, WC, but not BMI, showed a significant positive association with PTA-low (β ± s.e = 0.47 ± 0.10, P < 0.001, adjusted R2 = 0.2433) and PTA-high (β ± s.e = 0.70 ± 0.21, P = 0.001, adjusted R2 = 0.2006) in males younger than 55. In addition, WC, but not BMI, showed a borderline significant positive association with PTA-high in females older than 55 (β ± s.e. = 0.17 ± 0.09, P = 0.051, adjusted R2 = 0.2306). These findings suggest that WC is independently associated with ARHI, even after adjustment for BMI as a risk factor for ARHI.
The distribution of audiogram patterns by age, gender, and central obesity is shown in Table 4. The most common pattern was flat type for both nonobesity and obesity groups in female and male subjects younger than 55, and was abrupt high-tone loss type for both groups in male subjects older than 55. As for female subjects older than 55, the most common pattern was flat type for nonobesity group, but was abrupt high-tone loss type for obesity group. However, the audiogram patterns were not significantly associated with central obesity by any grouping (Table 4).
Table 4. Distribution of audiogram patterns by age, gender, and central obesity
This cross-sectional study provides novel findings regarding the possible association of central obesity with the severity of ARHI in adults over the age of 35. Our results, together with a report by Fransen et al. (15), suggest that being overweight or obese is associated with ARHI. Furthermore, our study suggests that WC is independently associated with ARHI, even after adjustment for BMI as a risk factor for ARHI. These results reinforce the reported observations that central obesity is more important than BMI as a risk factor for human health hazards, such as the abnormal lipid/lipoprotein profiles in women (20), metabolic syndrome (3,21,22,23), nonalcoholic fatty liver disease (24) and mortality among postmenopausal women with CAD (25), and mortality (10).
The association between WC and ARHI may differ by age and gender. Our study showed that WC was associated with the ability to hear high frequencies in female subjects older than 55 and with the ability to hear low and high frequencies in males younger than 55. These age- and gender-specific presentations of ARHI are very similar to that of heart disease, in which men usually develop heart disease earlier than women, but women rapidly catch up in incidence of heart disease following menopause (26). The underlying causes of these gender and age differences in ARHI are not yet clear. We speculated that they may be related to a gender-related chronological difference in the development of central obesity.
In a nationwide survey in Taiwan, the age-standardized prevalence of abdominal obesity increased with age in both genders (27). The prevalence of central and generalized obesity in men was higher than in women between the age of 30 and 50 and remained stable afterward. However, women rapidly caught up by the age of 50–59 years, and the prevalence became higher in men afterward. This rapid increase in the prevalence of obesity in women around age 50–59 and after may be the result of menopause. In addition, estrogen was shown to play a protective role in the function (28,29) and plasticity (30) of the auditory system in females. The early development of central obesity in men may explain the association of WC with both PTA-low and PTA-high in males younger than 55. In contrast, the later development and higher prevalence of central obesity may explain the association of only PTA-high in females over 55. However, compared to noise, WC has only a minor contribution to ARHI in subjects older than 55.
The underlying mechanisms connecting WC and ARHI are not very clear. Central obesity is a major cause of type 2 diabetes mellitus and is an important risk factor for metabolic syndrome, and prospective data show that metabolic syndrome increases the risk not only of coronary artery disease but also of cerebrovascular disease (3). For type 2 diabetes mellitus, Maia and Campos (31) concluded that the etiopathogenesis of hearing loss associated with diabetes might be due to angiopathy, neuropathy, or even a combination of both conditions. The possible biochemical pathways for diabetes-related auditory dysfunction might be though oxidative stress and deposit of advanced glycation end products (12). For cerebrovascular disease, it could cause hypoxia, imbalance of redox status, and subsequent mitochondrial dysfunction of the cochlea (32). Hypoxia of the cochlea may be the cause of the mtDNA4977 deletion and possibly other mitochondrial DNA mutants. These may cause a further reduction in mitochondrial oxidative phosphorylation and decreased function of the acoustic neural system (33). Regarding dyslipidemia, it was shown to cause profound edema in the strial marginal layer and slight edema in the outer hair cell, mainly in the basal turn, in guinea-pigs (34). In our study, we have excluded major neurological diseases, such as stroke. Therefore, the underlying mechanism connecting central obesity and ARHI may involve a complex network, including macro- and micro-angiopathy, imbalance of redox status, and neuropathy secondary to the metabolic derangements of obesity-induced systemic diseases. Even so, it is still unknown whether central obesity itself can directly lead to ARHI.
The adipose tissue is now considered to be an endocrine tissue. It secretes both hormones and cytokines, influencing appetite, insulin resistance, energy metabolism, and atherosclerosis (35). Other adipose secretory substances include fatty acids, tumor necrosis factor-α, interleukin-6, leptin, adiponectin, resistin, visfatin, omentin, and many others (36). In addition, obesity-induced inflammation also results in the infiltration of macrophages and the release of the cytokines tumor necrosis factor-α, interleukin-6, and interleukin-1β. High levels of nitric oxide and reactive nitrogen species may be subsequently generated via induction of inducible nitric oxide synthase (37). However, low plasma adiponectin was reported to be associated with all components of the metabolic syndrome and many other diseases (38). Therefore, it is reasonable to surmise that obesity may directly cause hearing loss via obesity-related oxidative stress. Further research is required to elucidate the potential mechanisms linking obesity and/or its related adipocytokines on hearing.
In conclusion, our study found that WC was an independent risk factor of ARHI, particularly for low and high frequencies in males younger than 55 and for high frequencies in female subjects older than 55. Central obesity was more important than BMI as a risk factor for ARHI.
This study was supported by funding from the National Science Council of Taiwan (NSC 96-2314-B-002-132) and from Buddhist Dalin Tzu Chi General Hospital (DTCRD-96 (2)-09).