Association Between Central and Peripheral Age-Related Hearing Loss and Different Frailty Phenotypes in an Older Population in Southern Italy

This cross-sectional study assesses whether physical frailty or cognitive frailty phenotypes are associated with age-related hearing loss in an older community-dwelling population in Southern Italy.

Key Points

Question

Are physical frailty or cognitive frailty phenotypes associated with age-related hearing loss in an older community-dwelling population?

Findings

In this population-based cross-sectional study of 1929 community-dwelling older individuals in southern Italy, the prevalence of peripheral age-related hearing loss and age-related central auditory processing disorder (CAPD) was higher in physical and cognitive frailty groups than in the nonfrail group. Age-related CAPD was associated only with cognitive frailty.

Meaning

In this study, age-related CAPD was independently associated with cognitive frailty, suggesting that management of age-related hearing loss may be associated with development of different frailty phenotypes.

Abstract

Importance

The association between age-related hearing loss (ARHL) and physical or cognitive frailty has been poorly explored. These associations could define new perspectives for delaying frailty-related processes in older age.

Objective

To examine whether peripheral ARHL and age-related central auditory processing disorder (CAPD) are independently associated with physical or cognitive frailty.

Design, Setting, and Participants

This cross-sectional study analyzed registry data from December 31, 2014, on 1929 older (≥65 years) participants of the Salus in Apulia Study (Southern Italy) who underwent audiologic, physical, and neuropsychological assessment. Data analysis was performed from December 12, 2019, to January 4, 2020.

Main Outcomes and Measures

Prevalence of peripheral ARHL in older individuals with physical and/or cognitive frailty and those without frailty assessed using the Fried criteria (physical) and the Mini-Mental State Examination (cognitive). Multivariable logistic regression models were used to assess associations of audiologic variables with frailty phenotype.

Results

Data from 1929 participants (mean [SD] age, 73.6 [6.3] years; 974 male [50.5%]) were eligible for the analyses. The prevalence of peripheral ARHL was higher in the physical frailty group (96 [26.6%]) than in the nonfrail group (329 [21.0%]) (difference, 5.61 percentage points; 95% CI, 0.63-10.59 percentage points) and in the cognitive frailty group (40 [38.8%]) than in the nonfrail group (385 [21.1%]) (difference, 17.75 percentage points; 95% CI, 8.2-27.3 percentage points). Age-related CAPD was more prevalent in the physical frailty group (62 [17.2%]) than in the nonfrail group (219 [14.0%]) (difference, 3.21 percentage points; 95% CI, −1.04 to 7.46 percentage points) and in the cognitive frailty group (28 [27.2%]) than in the nonfrail group (253 [13.9%]) (difference, 13.33 percentage points; 95% CI, 4.10-22.21 percentage points). In the multivariable models, age-related CAPD was associated with cognitive frailty in the fully adjusted model (odds ratio [OR], 1.889; 95% CI, 1.094-3.311). There was also an inverse association between the unitary increase in Synthetic Sentence Identification With the Ipsilateral Competitive Message scores, indicating a lower likelihood of this disorder, and cognitive frailty (OR, 0.989; 95% CI, 0.988-0.999). Peripheral ARHL was associated with cognitive frailty only in the partially adjusted model (OR, 1.725; 95% CI, 1.008-2.937).

Conclusions and Relevance

In this cross-sectional study of 1929 participants, age-related CAPD was independently associated with cognitive frailty. Whether the management of ARHL may help prevent the development of different frailty phenotypes or improve their clinical consequences should be addressed in longitudinal studies and, eventually, well-designed randomized clinical trials.

Introduction

Aging is the strongest risk factor for many chronic medical conditions. Such age-related susceptibility is thought to be caused by a progressive imbalance between the challenges associated with internal and external stressors (ie, sensory impairments, psychosocial stress, diseases, and injuries) and progressively failing resilience mechanisms, eventually leading to a break in the physiologic homeostasis that is clinically manifested as frailty, disability, and death.¹ Frailty has been associated with multiple adverse health-related outcomes, including falls, mobility loss, disability in activities of daily living, and early mortality.^2,3 During the past 3 decades, several operational definitions of frailty have been proposed. The most commonly used is the frailty phenotype considered as a medical syndrome and originally described by Fried et al⁴ using data from the Cardiovascular Health Study. The phenotypic definition of frailty by Fried et al⁴ is rooted in measures of physiologic and functional impairments, such as poor strength, impaired mobility, and unexplained weight loss. However, even though the underlying mechanisms of frailty are not fully understood, frailty is acknowledged to be not only a biological or physiologic state but also a multidimensional concept, including physical, cognitive, sensorial, social, psychological or depressive, and nutritional phenotypes.¹ The concept of frailty has been extended to include several aspects of age-related functional decline, in particular cognitive impairment.^5,6,7,8,9,10

The association between cognitive impairment and physical frailty has led some authors to propose a definition of cognitive frailty in which cognitive decline and high risk of developing dementia are the preeminent features.^1,11 Sensory impairments, particularly age-related hearing loss (ARHL), play an important role in the development of this syndrome.¹² After 65 years of age,¹³ ARHL or presbycusis is a common disease, with a prevalence that increases exponentially with aging, and almost 100% of the oldest individuals (>80 years of age) report some degree of hearing loss. The 2 main features of ARHL are peripheral ARHL and age-related central auditory processing disorder (CAPD). Peripheral ARHL is mostly related to a progressive decline in cochlear function, assessed by pure tone audiometry, whereas age-related CAPD refers to the central auditory pathway disorders that imply difficulties in understanding speech against background noise or competitive speech that cannot contribute to peripheral ARHL.¹⁴ Age-related hearing loss has been independently associated with poorer cognitive functioning, accelerated cognitive decline, incident dementia, falls, and a slower gait.¹² Similarly, the association between ARHL and cognitive impairment has been explored at the biological, clinical, and epidemiologic levels,^14,15 but whether ARHL is a frequent feature of physical and/or cognitive frailty is unknown. Understanding the association between the different features of ARHL and physical and cognitive frailty is important because proper ARHL management may be an effective strategy to prevent frailty or at least moderate its clinical evolution and consequences.¹⁶ In the current study, we conducted a cross-sectional analysis of the association of age-related CAPD and peripheral ARHL with the most widely investigated frailty phenotypes (physical and cognitive) in an older cohort from the Salus in Apulia Study.¹⁷

Methods

Study Population and Design

Data used in this cross-sectional study were from a study of a representative population of residents in Castellana Grotte (Puglia region, Southern Italy) who were 65 years or older at the time of baseline recruitment. The study design and data collection method are described in detail elsewhere.^17,18 In brief, the sampling frame was the 19 675 residents listed in the Apulian Regional Health Registry on December 31, 2014, of whom 4021 were 65 years or older. The initial study sample included 2038 participants, with a participation rate of 50.7%. A total of 98 participants were excluded because of a clinically confirmed diagnosis of dementia, and 11 were excluded because of middle ear diseases, leaving 1929 participants available for analysis. Data analysis was performed from December 12, 2019, to January 4, 2020. All participants provided written informed consent, and the study was approved by the institutional review board of the National Institute of Gastroenterology Saverio De Bellis, where all the examinations described in this study were performed. All data were deidentified. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Peripheral ARHL Assessment

Before the hearing assessments, all participants underwent otoscopic and tympanometric examination (A222 Impedentiometer, Interacoustics) to assess middle-ear diseases. Peripheral ARHL was assessed with pure tone audiometry, according to the modified Hughson-Westlake method,¹⁹ in a soundproof booth with HDR 39 headphones (Sennheiser Electronic GmbH & Co KG) and the Piano Audiometer (Inventis SRL), calibrated and executed according to international standards for audiometric testing.²⁰ Pure tone average (PTA) was calculated at the middle-low frequencies of 0.5, 1, and 2 KHz, using only the label PTA, and at the highest frequencies of 4, 6, and 8 KHz, using the label PTA high frequencies (PTA HF).²¹ Separate analysis of PTA and PTA HF was needed to explore the association with frailty phenotypes for the entire audiometric pattern, from 0.5 to 8 KHz. Because peripheral ARHL is usually referred to as a greater than 25 dB hearing level, overestimating the prevalence of this condition in a population-based sample of those 65 years and older, in the current study, peripheral ARHL was defined as a PTA threshold greater than 40 dB hearing level in the better ear according to the World Health Organization definition of disabling ARHL.¹³ The Speech Discrimination Score (SDS) was defined as the percentage of recognition of a list of 10 phonetically balanced Italian words at a 30 dB sensation level over the PTA threshold for each ear.²²

Age-Related CAPD Assessment

Age-related CAPD was assessed only in the participants without disabling ARHL (ie, PTA<40 dB hearing level in the better ear and a speech recognition score at 30 dB over the subjective level >70% in the better ear). The test used to diagnose age-related CAPD was the Italian version of the Synthetic Sentence Identification With Ipsilateral Competitive Message (SSI-ICM) test²³ to assess central auditory dichotic processing. The test consists of administering for each ear a primary signal of 10 short sentences against a background competition signal (a male talker reading a passage). We chose to present the short sentences at a hearing comfort level of 50 dB sensation level over the PTA for each ear, according to the method used by Cooper and Gates²⁴ in the Framingham cohort. The rate of identification of sentences is expressed as a proportion (0%-100%) at various primary-competitive ratios (0, +5, +10 dB sound pressure level).²³ In accordance with Gates and colleagues,^25,26 age-related CAPD was considered present when the patient scored less than 50% in the better ear with a 0-dB message-competition ratio.

Assessment of Frailty Phenotypes

To assess the physical frailty status, we used the operational definition of Fried et al,⁴ namely, the positivity of 3 or more of the following criteria: weight loss, exhaustion, low levels of physical activity, weakness, and slowness (slightly modified for the current study). The 5-repetitions sit-to-stand test measures the amount of time a patient takes to rise 5 times from a seated position without using his or her arms and was used as a proxy measure of weakness using more than 15 seconds as a diagnostic threshold.^27,28 Nutritional status was assessed using the Mini-Nutritional Assessment Scale, which provides information on weight loss and nutritional intake using a score threshold of less than 23.5.²⁹ Gait speed was evaluated using a 5-m walking test and rated slow if the time recorded was greater than or equal to the cutoff point of 0.6 m/s. Physical activity was assessed by means of an interviewer-administered questionnaire (cutoff value <3 of 6 items that identified the intensity and frequency of physical activity in the past year).^30,31 The step test was used as a measure of exhaustion and assessed using a modified version of the Berg stool-stepping task.³² In the current study, we chose not to use the intermediate dimension of prefrailty status⁴ and to consider prefrail participants together with healthy participants. The cognitive frailty diagnosis was made according to the International Academy on Nutrition and Aging and the International Association of Gerontology and Geriatrics consensus criteria³³ as a combination of the presence of physical frailty and a Mini-Mental State Examination (MMSE) score of less than 24.³⁴

Other Covariates

For each individual, a blood sample was collected in the morning after overnight fasting to measure the levels of fasting blood glucose (FBG), total cholesterol, high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C). The FBG level was measured using the glucose oxidase method (Sclavus), whereas the concentrations of plasma lipids (triglycerides, total cholesterol, and HDL-C) were quantified with an automated colorimetric device (Hitachi, Boehringer Mannheim). The LDL-C levels were measured using the Friedewald equation. Cognitive mental status was assessed with the MMSE, which includes 11 questions and concentrates only on the global cognitive aspects of mental functions.³⁴ The clinical evaluation included extemporaneous ambulatory systolic blood pressure (SBP) and diastolic blood pressure (DBP), determined with the patient in a sitting position after at least a 10-minute rest and taken at least 3 different times using an automatic blood pressure monitor (Omron Healthcare). The diagnoses of myocardial infarction and stroke were made based on the presence of diagnosis of these diseases as revealed in medical records and general practitioner information and/or use of medications for these diseases.¹⁸ Diabetes and hypertension were diagnosed based on the following international diagnostic standards: FBG level higher than 125 g/dL (to convert to millimoles per liter, multiply by 0.0555) and SBP/DBP greater than or equal to 130/80 mm Hg. Finally, the body mass index was calculated as weight in kilograms divided by height in meters squared.

Statistical Analysis

The whole sample was categorized according to the frailty phenotypes to describe clinical and functional differences in terms of frequency and associations. In addition, to assess the clinical differences between different hearing loss exposures, we categorized the study sample as individuals without disabling ARHL, individuals with age-related CAPD, and individuals with peripheral ARHL. Because of the nonnormal distribution of the variables, continuous variables were expressed as median (interquartile range [IQR]) and categorical variables as number (percentage). Differences in the prevalence of outcomes (frailty phenotypes), exposure groups (hearing loss), and other categorical variables and their 95% CIs were calculated and used to assess practically important differences in terms of the magnitude of association (effect sizes). Median of the difference of medians between the groups (outcome vs nonoutcome and exposure vs nonexposure) was considered as the effect size for continuous variables. The 95% CI around that difference was calculated following the method of Altman et al.³⁵ Binary logistic regression models were used to assess associations of audiologic variables (age-related CAPD, peripheral ARHL, SSI-ICM, PTA, PTA HF, and SDS) with frailty phenotype (physical frailty and cognitive frailty) as dependent variables.

We built 3 nested models in which lower levels were sorted under a hierarchy of successively higher-level units (in terms of covariate adjustment). The first was an unadjusted model with just the audiologic diagnosis (peripheral ARHL and age-related CAPD) as independent variables, a partially adjusted model corrected only for age and sex to assess their effect on the association regardless of the other major confounders, and a fully adjusted model corrected for age, sex, educational level, hypertension, diabetes, stroke, myocardial infarction, total cholesterol level, FBG level, SBP, DBP, and smoking status. All model assumptions were tested for collinearity using a correlation matrix and then the variance inflation factor (VIF) test for all the logistic regression models.³⁶ Collinearity of all variables used in the models was assessed using a VIF score greater than 2.5. However, none of these variables exceeded this cut-off, and for this reason they have all been maintained in the models. The audiologic variables used as covariates were dichotomous (age-related CAPD and peripheral ARHL) and continuous (SSI-ICM, PTA, PTA HF, and SDS). In the fully adjusted models with cognitive and physical frailty as dependent variables and age-related CAPD and SSI-ICM as independent variables, we also added PTA among the covariates’ adjustment to exclude the residual confounding effect of peripheral ARHL (measured by PTA) in the association. In addition, in the SSI-ICM model as an independent variable to test the interaction between peripheral hearing function measured by PTA and the association between SSI-ICM and cognitive and physical phenotype, we used the interaction factor SSI-ICM × PTA as an adjustment covariate. Individuals with peripheral ARHL were excluded from all analyses that used SSI-ICM scores. Statistical analysis was performed using R, version 1.2.5042 (R Project for Statistical Computing) using the following expansion packages: Tidyverse (data management), kableExtra and gmodels (tabulated outputs), car (VIF collinearity test), ggplot2 and ggthemes (for graphics), and epiR (prevalence proportions and relative 95% CIs). For effect size calculations and relative 95% CIs, we used Stata, version 16.0 (StataCorp LLC).

Results

Descriptive Analyses

Overall, 1929 participants (mean [SD] age, 73.6 [6.3] years; 974 male [50.5%]) were eligible for the analyses presented in this study. The mean (SD) educational level of the participants was 6.9 (3.8) years, and the mean (SD) MMSE score was 26.3 (4.3). The overall prevalence of physical frailty was 18.7% (95% CI, 17.0%-20.5%). Moreover, the prevalence of physical frailty was higher among women (20.8% [199 of 955]) than among men (16.6% [162 of 974]), with a difference of 4.2 percentage points (95% CI, 0.7-7.7 percentage points). The overall prevalence of cognitive frailty was 5.3% (95% CI, 4.4%-6.4%), with no meaningful differences in sex between cognitively frail and nonfrail groups, although 59 of the 955 women (6.2%) were assigned to the cognitive frailty group vs 44 of the 974 men (4.5%) (difference, 1.8 percentage points; 95% CI, −0.3 to 3.7 percentage points). The prevalence of age-related CAPD was 14.6% (281 of 1929), and the prevalence of peripheral ARHL was 22.0% (425 of 1929).

Table 1 compares sociodemographic and clinical variables between physically frail and nonfrail participants. Physically frail participants were older and less educated and more likely to be female. Individuals with peripheral ARHL were more often found in the physical frailty group (96 [26.6%]) than in the nonfrail group (329 [21.0%]), with a prevalence difference of 5.61 percentage points (95% CI, 0.63-10.59 percentage points). Age-related CAPD was more common in the physical frailty group (62 [17.2%]) than in the nonfrail group (219 [14.0%]), with a prevalence difference of 3.21 percentage points (95% CI, −1.04 to 7.46 percentage points). Participants with physical frailty had a higher PTA HF (median, 62.5 dB hearing level; IQR, 20.0-120.0 dB hearing level; effect size, 2.5; 95% CI, 0-5.0) and a lower SSI-ICM score (median, 65; IQR, 0 to 100; effect size, −5; 95% CI, −10 to 0) compared with those in the nonfrail group. The prevalence of diabetes was higher in physically frail individuals compared with those in the nonfrail group (Table 1), with a proportion difference of 4.8 percentage points (95% CI, 0.7-9.0 percentage points).

Table 1. Sociodemographic and Clinical Variables Among 1929 Participants With and Without Physical Frailty in the Salus in Apulia Study^a.

Variable	Individuals with physical frailty (n = 361)	Individuals without physical frailty (n = 1568)	Effect size (95% CI)b
Sociodemographic
Age, y	75 (65-95)	72 (65-95)	2 (1 to 3)
Sex
Female	199 (55.1)	756 (48.2)	6.91 (1.21 to 12.61)
Male	162 (44.9)	812 (51.8)	−6.91 (−12.61 to −1.21)
Educational level, y	5 (0 to 17)	5 (0 to 23)	−1 (−1 to 0)
Smoked	19 (5.3)	132 (8.4)	−3.16 (−5.16 to −0.03)
BMI	28.5 (17.4 to 48.8)	28 (14.09 to 52.45)	0.27 (−0.28 to 0.83)
MMSE score	27 (5 to 30)	28 (1 to 30)	−1 (−1 to −1)
Hearing measurement
Age-related CAPD	62 (17.2)	219 (14.0)	3.21 (−1.04 to 7.46)
Peripheral ARHL	96 (26.6)	329 (21.0)	5.61 (0.63 to 10.59)
PTA, dB HL	30.0 (13.5 to 95.0)	30.0 (12.5 to 107.5)	0 (0 to 2.5)
PTA HF, dB HL	62.5 (20.0 to 120.0)	60.0 (15.0 to 122.5)	2.5 (0 to 5.0)
SDS, %	95 (0 to 100)	100 (0 to 100)	0 (0 to 0)
SSI-ICM score, %	65 (0 to 100)	75 (0 to 100)	−5 (−10 to 0)
Metabolic biomarker
FBG level, mg/dL	99 (54 to 435)	99 (59 to 365)	1 (−1 to 3)
Diabetes	60 (16.6)	185 (11.8)	4.82 (0.66 to 8.98)
Total cholesterol level, mg/dL	181 (79 to 302)	183.5 (20 to 386)	−3 (−8 to 1)
HDL-C level, mg/dL	46 (21 to 105)	46.5 (22 to 495)	−1 (−2 to 1)
LDL-C level, mg/dL	111 (28 to 217)	113 (14 to 220)	−3 (−6 to 1)
Triglyceride levels, mg/dL	94 (21 to 506)	93 (6 to 773)	2 (−3 to 7)
SBP, mm Hg	130 (100 to 200)	130 (80 to 180)	0 (0 to 0)
DBP, mm Hg	80 (50 to 140)	80 (40 to 110)	0 (0 to 0)
Clinical
Hypertension	309 (85.6)	1326 (84.6)	1.03 (−3.01 to 5.07)
Myocardial infarction	16 (5.0)	72 (5.2)	−0.16 (−2.83 to 2.51)
Stroke	10 (3.1)	31 (2.2)	0.91 (−1.16 to 2.98)
Weakness	334 (92.5)	272 (17.3)	75.17 (71.88 to 78.47)
Exhaustion	180 (49.9)	48 (3.1)	46.80 (41.57 to 52.03)
Slowness	317 (87.8)	177 (11.3)	76.52 (72.80 to 80.24)
Weight loss	54 (15.0)	75 (4.8)	10.18 (6.35 to 14.00)
Low physical activity	356 (98.6)	1243 (79.3)	19.34 (17.00 to 21.68)

Abbreviations: ARHL, age-related hearing loss; BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); CAPD, central auditory processing disorder; DBP, diastolic blood pressure; FBG, fasting blood glucose; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; MMSE, Mini-Mental State Examination; PTA, pure tone average; PTA HF, PTA high frequencies; SBP, systolic blood pressure; SDS, Speech Discrimination Score; SSI-ICM, Synthetic Sentence Identification With Ipsilateral Competitive Message.

SI conversion factors: to convert glucose to millimoles per liter, multiply by 0.0555; total cholesterol, HDL-C, and LDL-C to millimoles per liter, multiply by 0.0259; and triglycerides to millimoles per liter, multiply by 0.0113.

^a Data are presented as median (interquartile range) for continuous variables and as number (percentage) for categorical variables. The Salus in Apulia Study is reported in Sardone et al.¹⁷

^b For continuous variables, effect size is the median of the difference of medians between the groups, and the 95% CI around that difference was calculated using the method of Altman et al.³⁵ For categorical variables, effect size is the difference between proportions of the groups with relative 95% CIs calculated around that difference.

Table 2 compares sociodemographic and clinical variables between individuals with cognitive frailty and nonfrail participants. Cognitively frail participants were older, less educated, and more likely to be female than were nonfrail participants. Within the cognitive frailty group, the prevalence of age-related CAPD was 27.2% (28 of 103), and the prevalence of peripheral ARHL was 38.8% (40 of 103); these proportions were higher than in the nonfrail group (difference, 13.33 percentage points [95% CI, 4.59-22.07 percentage points] for age-related CAPD and 17.75 percentage points [95% CI, 8.15-27.35 percentage points] for peripheral ARHL). All the hearing parameters, such as PTA, PTA HF, SDS, and SSI-ICM, suggested a poorer audiometric performance in cognitively frail individuals compared with nonfrail participants. All meaningful differences found when comparing the physical frailty and nonfrailty groups were substantially replicated when cognitively frail individuals were compared with nonfrail participants (Table 1 and Table 2) with the exception of total cholesterol, HDL-C, and LDL-C, which were higher in the nonfrailty group than in the cognitive frailty group. Many of the differences found between the physical and cognitive frailty groups and the nonfrailty group may have been associated with the substantial age difference between these groups. Participants with ARHL had worse performance on physical frailty components (gait speed, exhaustion, and weakness) with the exception of physical activity, which was lower in the peripheral ARHL and age-related CAPD groups than in those without disabling peripheral ARHL (Table 3).

Table 2. Sociodemographic and Clinical Variables Among 1929 Participants With and Without Cognitive Frailty in the Salus in Apulia Study^a.

Variable	Individuals with cognitive frailty (n = 103)	Individuals without cognitive frailty (n = 1826)	Effect size (95% CI)b
Sociodemographic
Age, y	79 (66 to 95)	72 (65 to 95)	6 (4 to 7)
Sex
Female	59 (57.3)	896 (49.10)	8.21 (−1.61 to 18.04)
Male	44 (42.7)	930 (50.90)	−8.21 (−18.04 to 1.61)
Educational level, y	3 (0 to 13)	5 (0 to 23)	−3 (−3 to −2)
Smoked	4 (3.9)	147 (8.1)	−4.17 (−8.10 to 0.23)
BMI	28.6 (17.4 to 48.8)	28.0 (14.1 to 52.4)	0.58 (−0.39 to 1.61)
MMSE score	20 (5 to 23)	28 (1 to 30)	−8 (−8 to −7)
Hearing measurement
Age-related CAPD	28 (27.2)	253 (13.9)	13.33 (4.59 to 22.07)
Peripheral ARHL	40 (38.8)	385 (21.1)	17.75 (8.15 to 27.35)
PTA, dB HL	35 (17.5 to 95.0)	30 (12.5 to 107.5)	5 (2.5 to 7.5)
PTA HF, dB HL	70 (22.5 to 120.0)	60 (15.0 to 122.5)	10 (7.5 to 15)
SDS, %	90 (0 to 100)	100 (0 to 100)	−5 (−10 to −5)
SSI-ICM score, %	35 (0 to 100)	75 (0 to 100)	−30 (−40 to −25)
Metabolic biomarker
FBG level, mg/dL	101 (78 to 349)	99 (54 to 435)	3 (0 to 7)
Diabetes	24 (23.3)	221 (12.1)	11.20 (2.90 to 19.50)
Total cholesterol level, mg/dL	175 (94 to 283)	183 (20 to 386)	−10 (−18 to −2)
HDL-C level, mg/dL	44 (21 to 105)	46 (22 to 495)	−2 (−5 to 0)
LDL-C level, mg/dL	106 (37 to 199)	113 (14 to 220)	−8 (−15 to −1)
Triglyceride levels, mg/dL	98 (38 to 323)	93 (6 to 773)	5 (−3 to 14)
SBP, mm Hg	130 (100 to 170)	130 (80 to 200)	0 (0 to 0)
DBP, mm Hg	80 (50 to 95)	80 (40 to 140)	0 (0 to 0)
Clinical
Hypertension	84 (81.6)	1551 (84.9)	−3.39 (−11.05 to 4.28)
Myocardial infarction	6 (6.2)	82 (5.1)	1.15 (−3.81 to 6.11)
Stroke	4 (4.2)	37 (2.3)	1.87 (−2.20 to 5.93)
Weakness	96 (93.2)	510 (27.9)	65.27 (60.00 to 70.55)
Exhaustion	53 (51.0)	175 (9.6)	41.87 (32.13 to 51.62)
Slowness	93 (90.3)	401 (22.0)	68.33 (62.31 to 74.36)
Weight loss	15 (14.6)	114 (6.2)	8.32 (1.42 to 15.22)
Low physical activity	101 (98.1)	1498 (82.0)	16.02 (12.83 to 19.21)

Abbreviations: ARHL, age-related hearing loss; BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); CAPD, central auditory processing disorder; DBP, diastolic blood pressure; FBG, fasting blood glucose; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; MMSE, Mini-Mental State Examination; PTA, pure tone average; PTA HF, PTA high frequencies; SBP, systolic blood pressure; SDS, Speech Discrimination Score; SSI-ICM, Synthetic Sentence Identification With Ipsilateral Competitive Message.

SI conversion factors: to convert glucose to millimoles per liter, multiply by 0.0555; total cholesterol, HDL-C, and LDL-C to millimoles per liter, multiply by 0.0259; and triglycerides to millimoles per liter, multiply by 0.0113.

^a Data are presented as median (interquartile range) for continuous variables and as number (percentage) for categorical variables. The Salus in Apulia Study is reported in Sardone et al.¹⁷

^b For continuous variables, effect size is the median of the difference of medians between the groups, and the 95% CI around that difference was calculated using the method of Altman et al.³⁵ For categorical variables, effect size is the difference between proportions of the groups with relative 95% CIs calculated around that difference.

Table 3. Sociodemographic and Clinical Variables Among 1929 Participants With and Without Different Age-Related Hearing Loss Subtypes in the Salus in Apulia Study^a.

Variable	Individuals with peripheral ARHL (n = 425)	Individuals with age-related CAPD (n = 281)	Individuals without disabling ARHL (n = 1223)	Effect size (95% CI)b
				Peripheral ARHL vs without disabling ARHL	Age-related CAPD vs without disabling ARHL
Sociodemographic
Age, y	77 (65 to 95)	76 (66 to 95)	71 (65 to 90)	5 (4 to 6)	4 (3 to 5)
Female	229 (53.9)	141 (50.2)	585 (47.8)	−6.05 (−11.55 to −0.54)	−2.34 (−8.83 to 4.14)
Educational level, y	5 (0 to 23)	5 (0 to 18)	6 (0 to 18)	−1 (−1 to 0)	−1 (−2 to 0)
Smoked	26 (6.1)	15 (5.3)	110 (9.0)	−2.88 (−5.66 to −0.09)	−3.66 (−6.73 to −0.58)
BMI	28.2 (17.4 to 47.6)	28.8 (19.1 to 48.8)	27.8 (14.1 to 52.4)	0.12 (−0.41 to 0.60)	0.76 (0.05 to 1.42)
MMSE score	26 (5 to 30)	25 (1 to 30)	28 (2 to 30)	−2 (−2 to −1)	−2 (−3 to −2)
Hearing measurement
PTA, dB HL	50.0 (40.0 to 107.5)	32.5 (17.5 to 77.5)	25.0 (12.5 to 60.0)	25.0 (25.0 to 25.0)	7.5 (5.0 to 7.5)
PTA HF, dB HL	80.0 (20.0 to 120.0)	67.5 (22.5 to 122.5)	50.0 (15.0 to 110.0)	30.0 (27.5 to 30.8)	15.0 (12.0 to 17.0)
SDS	80 (0 to 145)	90 (0 to 100)	100 (30 to 100)	−15 (−15 to −15)	−5 (−10 to −5)
SSI-ICM score	20 (0 to 100)	25 (0 to 70)	90 (50 to 100)	−60 (−65 to −55)	−60 (−65 to −60)
Metabolic biomarker
FBG level, mg/dL	100 (73 to 435)	100 (70 to 349)	99 (54 to 355)	2 (0 to 3)	1 (−1 to 3)
Diabetes	69 (16.2)	47 (16.7)	129 (10.5)	5.69 (1.78 to 9.59)	6.18 (1.49 to 10.87)
Total cholesterol level, mg/dL	181 (95 to 287)	175 (76 to 386)	185 (20 to 386)	−5 (−9 to −1)	−10 (−15 to −5)
HDL-C level, mg/dL	45 (23 to 495)	46 (21 to 265)	47 (22 to 106)	−2 (−3 to 0)	−1 (−3 to 0)
LDL-C level, mg/dL	111 (14 to 213)	105 (28 to 195)	114 (30 to 220)	−4 (−7 to 0)	−10 (−14 to −5)
Triglyceride levels, mg/dL	95 (21 to 483)	86 (6 to 270)	93 (23 to 773)	0 (−5 to 5)	−4 (−10 to 1)
SBP, mm Hg	130 (100 to 180)	130 (100 to 180)	130 (80 to 200)	0	0
DBP, mm Hg	80 (40 to 110)	80 (60 to 140)	80 (50 to 100)	0	0
Clinical
Hypertension	346 (81.4)	237 (84.3)	1052 (86.0)	−4.61 (−8.78 to −0.43)	−1.68 (−6.35 to 3.00)
Myocardial infarction	18 (4.7)	15 (6.4)	55 (5.0)	−0.30 (−2.80 to 2.20)	1.40 (−2.01 to 4.80)
Stroke	10 (2.6)	11 (4.7)	20 (1.8)	0.80 (−0.99 to 2.60)	2.89 (0.05 to 5.73)
Weakness	136 (32.0)	92 (32.7)	378 (30.9)	1.09 (−4.04 to 6.23)	1.83 (−4.23 to 7.90)
Exhaustion	63 (14.3)	31 (11.0)	134 (11.0)	3.87 (0.06 to 7.67)	0.08 (−3.98 to 4.14)
Slowness	123 (28.9)	77 (27.4)	294 (24.0)	4.90 (−0.03 to 9.83)	3.36 (−2.38 to 9.10)
Weight loss	32 (7.5)	23 (8.2)	74 (6.1)	1.10 (−1.68 to 3.87)	2.13 (−1.34 to 5.61)
Low physical activity	360 (84.7)	243 (86.5)	996 (81.4)	3.27 (−0.79 to 7.32)	5.04 (0.48 to 9.59)
Physical frailty	62 (22.1)	96 (22.6)	203 (16.6)	5.99 (1.50 to 10.48)	5.47 (0.19 to 10.74)
Cognitive frailty	25 (5.9)	17 (6.0)	36 (2.9)	6.55 (3.62 to 9.48)	7.10 (3.48 to 10.73)

Abbreviations: ARHL, age-related hearing loss; BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); CAPD, central auditory processing disorder; DBP, diastolic blood pressure; FBG, fasting blood glucose; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; MMSE, Mini-Mental State Examination; PTA, pure tone average; PTA HF, PTA high frequencies; SBP, systolic blood pressure; SDS, Speech Discrimination Score; SSI-ICM, Synthetic Sentence Identification With Ipsilateral Competitive Message.

SI conversion factors: to convert glucose to millimoles per liter, multiply by 0.0555; total cholesterol, HDL-C, and LDL-C to millimoles per liter, multiply by 0.0259; and triglycerides to millimoles per liter, multiply by 0.0113.

^a Data are presented as median (interquartile range) for continuous variables and as number (percentage) for categorical variables. The Salus in Apulia Study is reported in Sardone et al.¹⁷

^b For continuous variables, effect size is the median of the difference of medians between the groups, and the 95% CI around that difference was calculated using the method of Altman et al.³⁵ For categorical variables, effect size is considered as the difference between proportions of the groups with relative 95% CIs calculated around that difference.

Multivariable Association Models

Table 4 details the logistic regression models used to measure the association between ARHL and frailty phenotypes. Both ARHL subtypes (peripheral ARHL and age-related CAPD) in terms of continuous and categorical variables were not associated with physical frailty in the adjusted models. In all models with physical frailty as the dependent variable, after adjusting for age and/or other covariates, the magnitude of the association was diminished (Table 4). When the same analyses were repeated with cognitive frailty as the outcome, the results were substantially different. Peripheral ARHL was associated with a nearly double increase in cognitive frailty in the partially adjusted model (OR, 1.725; 95% CI, 1.008-2.937) compared with lack of peripheral ARHL. When adjusted for all confounders (fully adjusted model), particularly educational level, the association decreased in magnitude by 33% (OR, 1.399; 95% CI, 0.805-2.415), but the difference was not significant. However, the probability of observing cognitive frailty among the participants with age-related CAPD compared with those without disabling peripheral ARHL was 2.5 times higher also when adjusted for age and sex (partially adjusted model) (OR, 2.604; 95% CI, 1.496-4.548). Furthermore, when corrected for all covariates, including educational level and PTA (fully adjusted model), the association remained (OR, 1.889; 95% CI, 1.094-3.311). The likelihood of observing cognitive frailty decreased by 2% for every unitary increase of SSI-ICM score regardless of the peripheral hearing threshold and other major confounders (fully adjusted model) (OR, 0.989; 95% CI, 0.988-0.999).

Table 4. Logistic Regression Models With Frailty Phenotypes as Dependent Variables and Regressors in the Salus in Apulia Study^a.

Variable	Odds ratio (95% CI)
	Unadjusted model	Partially adjusted modelb	Fully adjusted modelc
Physical frailty
Peripheral ARHL	1.472 (1.116-1.925)	1.071 (0.795-1.455)	1.048 (0.773-1.421)
Age-related CAPD	1.421 (1.033-1.958)	1.172 (0.835-1.646)	1.123 (0.785-1.681)d
PTA	1.012 (0.997-1.014)	1.003 (0.992-1.008)	1.006 (0.990-1.021)
PTA HF	0.981 (0.978-0.999)	1.001 (0.994-1.007)	0.999 (0.990-1.007)
SDS	1.000 (0.991-1.002)	1.001 (0.996-1.007)	1.002 (0.997-1.007)
SSI-ICM score	0.992 (0.989-0.997)	0.996 (0.993-1.000)	0.999 (0.996-1.002)e
SSI-ICM × PTA	0.986 (0.977-0.998)	0.999 (0.998-1.000)	0.999 (0.999-1.000)
Cognitive frailty
Peripheral ARHL	3.536 (2.208-5.631)	1.725 (1.008-2.937)	1.399 (0.805-2.415)
Age-related CAPD	3.769 (2.244-6.287)	2.604 (1.496-4.548)	1.889 (1.094-3.311)d
PTA	1.034 (1.017-1.040)	1.000 (0.991-1.016)	0.997 (0.968-1.022)
PTA HF	0.987 (0.997-0.999)	1.010 (1.000-1.023)	1.008 (0.989-1.021)
SDS	0.980 (0.973-0.988)	0.994 (0.984-1.001)	0.991 (0.987-1.005)
SSI-ICM score	0.980 (0.970-0.984)	0.986 (0.979-0.992)	0.989 (0.988-0.999)e
SSI-ICM × PTA	NA	NA	0.994 (0.987-1.000)

Abbreviations: ARHL, age-related hearing loss; CAPD, central auditory processing disorder; NA, not applicable; PTA, pure tone average; PTA HF, PTA high frequencies; SDS, Speech Discrimination Score; SSI-ICM, Synthetic Sentence Identification With Ipsilateral Competitive Message.

^a The Salus in Apulia Study is reported in Sardone et al.¹⁷

^b Adjusted for sex and age.

^c Adjusted for sex, age, educational level, hypertension, diabetes, stroke, myocardial infarction, total cholesterol level, fasting blood glucose level, systolic blood pressure, diastolic blood pressure, and smoking status.

^d Additionally adjusted for PTA.

^e Additionally adjusted for PTA and interaction covariate SSI-ICM × PTA (data shown in the last row).

Discussion

In the current study, the most important finding was the association of age-related CAPD with cognitive frailty, even when adjusted for age and a number of potential confounders. These findings were cross-validated by demonstrating an inverse linear association of SSI-ICM score, a continuous variable used to diagnose age-related CAPD, with cognitive frailty. In the fully adjusted model, for each unit of proportion increase in SSI-ICM, indicating a lower likelihood of age-related CAPD, there was a 2% decrease in the probability of having cognitive frailty. To our knowledge, no other studies have explored this association. When adjusted for all major confounders, age-related CAPD was not associated with physical frailty. Furthermore, peripheral ARHL was associated with cognitive frailty only in the partially adjusted model independently of age, whereas in the fully adjusted model the association was not maintained after adjustment for all other confounders, particularly educational level. No further meaningful association of peripheral ARHL with any frailty phenotype was found.

The prevalence estimate of 18.7% of physical frailty in a representative cohort of individuals 65 years or older is consistent with the 17% mean global prevalence in the population of Europe older than 65 years.³⁷ The prevalence of cognitive frailty according to the International Academy on Nutrition and Aging and the International Association of Gerontology and Geriatrics consensus criteria³³ was 5.3%. This frequency seems to be in line with another cross-sectional study³⁸ that described cognitive frailty in an older Italian population (≥65 years of age) using the same methods, finding a global prevalence of 4.4%. However, a systematic review¹ suggested considerable heterogeneity, with prevalence estimates ranging from 1.0% to 22.0% (10.7%-22.0% in clinical-based settings and 1.0%-4.4% in population-based settings). Moreover, the age prevalence curves of the frailty phenotypes in the current study revealed a robust increment of proportions among the age classes.

In the current study, physically and cognitively frail participants had worse auditory parameters than those in the nonfrail group. However, many of the differences found between the physical and cognitive frailty groups and the nonfrailty group may have been associated with the substantial age difference between these groups. This finding is consistent with the few studies^39,40,41,42 that investigated similar associations, although with a different, simplified method and without detecting age-related CAPD. The Korean Frailty and Aging Cohort Study⁴¹ found a positive association only between social frailty (social isolation plus physical frailty) and peripheral ARHL, with the social frailty group being older than the nonfrailty group. The English Longitudinal Study of Aging⁴⁰ found cross-sectional and longitudinal associations between physical frail status and poor hearing using a single subjective question. The Health, Aging and Body Composition Study³⁹ found an increased risk of developing physical frailty, detected only with the coexistence of a slow gait and inability to stand up from a chair, in individuals with peripheral ARHL. In addition, the National Health and Nutrition Examination Survey⁴² found that greater peripheral ARHL was independently associated with a slower gait speed and impaired motor domains of physical frailty. This finding was supported by the results of the current study that each component of the physical frailty phenotype by Fried et al,⁴ in particular a higher prevalence of low physical activity, was more prevalent in the peripheral ARHL and age-related CAPD groups than in normal-hearing individuals.

Some neurobiological factors that may be associated with ARHL, physical decline, and cognitive impairment have been hypothesized, including inflammatory processes and primitive neurodegeneration of the auditory cortex, together with nutritional, vascular, neuropathologic, and metabolic factors.^43,44 In the current study, age-related CAPD was independently associated with cognitive frailty but not with physical frailty, suggesting that central auditory impairments rather than peripheral components may play a role in accelerated cognitive performance impairments associated with aging. This finding is consistent with those of other population-based studies.^25,45,46 However, in the partially adjusted model (ie, only sex and age), peripheral ARHL was associated with cognitive frailty, whereas this association was not present in the fully adjusted model. In particular, educational level dominated the association in the fully adjusted model. One explanation of this particular finding is that educational level is a typical proxy determinant for cognitive reserve, which has previously been described as associated with both peripheral ARHL and cognitive impairment.^47,48 The confounding effect of educational level may explain the weak association in the partially adjusted model. Because the diagnosis of CAPD is impossible in individuals with advanced deterioration of the cochlea, many individuals with age-related CAPD masked by the peripheral ARHL may have been lost to follow-up. For this reason, in the fully adjusted models with age-related CAPD as a covariate, the effect of peripheral ARHL using the PTA (expression of the peripheral hearing function) as a covariate was also tested. In addition, the interaction factor SSI-ICM × PTA as a covariate was used to test the possible interaction of peripheral ARHL in the association between the unitary increase of SSI-ICM score and both cognitive and physical frailty. However, there were no changes in the strength of the association. Cognitive frailty was associated only with SSI-ICM score in the fully adjusted model, and SSI-ICM × PTA was no longer associated with cognitive frailty. This latter result suggests that PTA and peripheral hearing status were not associated with the SSI-ICM score.

The present findings support the suggestive theory that the association between age-related CAPD and cognition might be related to the age-related degeneration of particular brain areas involved in the same cognitive functions used in the recognition of semantic and syntactic constructs against background noise or during competitive speech.¹⁵ The performances on central auditory tests used to assess central auditory processing, such as SSI-ICM in the current study, are associated with attentive-executive functions, short-term memory, task shifting, and attention to task.⁴⁹ The association between global cognitive functions, usually measured with MMSE,²⁵ and the current data (cognitive frailty was associated with a low MMSE score) supports the premise that central auditory testing could be regarded in part as a measure of cognitive function. Gates et al⁵⁰ demonstrated this hypothesis by adding central auditory test results to executive function scores. The degree of variability in cognitive impairment diagnoses increased, suggesting that central auditory assessment may help with diagnosis of cognitive impairment and dementia beyond the executive functioning measures.⁴

The association between age-related CAPD and physical frailty was not present in the partially and fully adjusted models including age as a covariate. This finding may have been attributable to the mechanisms underlying frailty phenotypes being mostly determined by aging processes. Moreover, frailty has been considered as a proxy category for accelerated aging and a precursor of mortality.⁷ Age also affected the association between peripheral ARHL and physical frailty; the magnitude of the association was smaller when adjusted for age in all models. Therefore, it is difficult to disentangle age from the association of ARHL with physical frailty and to express hypotheses based on biological evidence. However, the increased prevalence of ARHL in frailty phenotypes may be associated with inflammation processes. Recently, ARHL has been associated with inflammation-inducing dietary habits.^17,51 Chronic inflammation may also alter brain neurotropism, especially with an impaired expression of brain-derived neurotrophic factor,⁵² an important mediator of neurotropism both in the lower parts of the auditory cortex (also involved in CAPD neuropathologic mechanisms) and in the hippocampus (the most important location for memory).⁵³ Although systemic inflammation is considered one of the most prominent factors associated with cognitive decline and frailty,⁵⁴ the hypothesis of a similar association with ARHL is plausible, even if only a small body of evidence is yet available.^55,56,57 Furthermore, several mechanisms could explain the increased prevalence in our study of low physical activity among hearing impaired groups. Individuals with peripheral ARHL and age-related CAPD may have low levels of physical activity⁵⁸ partly because of social isolation⁵⁹ and an increased cognitive load.⁶⁰ Furthermore, attentive-executive functions^45,61 are important features for posture and balance.⁶² If we consider CAPD as a form of cognitive-related hearing impairment, it could impact the individual’s ability to effectively monitor the auditory environment, especially in crowded and noisy places, and thereby affect the individual’s likelihood of being able to perform physical activities.

Physical frailty and cognitive frailty are 2 major conditions that expose affected individuals to increases in dementia, institutionalization, and mortality. Peripheral ARHL has been reported to be the most important potentially modifiable factor associated with dementia,⁶³ and the current findings highlight the possibility that age-related CAPD may be associated with frailty. In clinical practice, frail, older patients with cognitive impairment and hearing difficulties, when in a noisy environment or against competitive speech, should also be tested for age-related CAPD. The findings suggest that central auditory testing may need to become a critical part of the comprehensive geriatric assessment.⁶⁴

Strengths and Limitations

Strengths of the study include that the measures used to assess physical and cognitive frailty and auditory dysfunction at peripheral and central level were standardized and also used in clinical settings. There was also good internal validity because the physical and cognitive biomarkers and auditory parameters were worse in the physically and cognitively frail groups. In addition, our prevalence data were consistent with the findings reported in similar population-based studies.^1,37,38

This study also has limitations. It was impossible to define the direction of the association because of the cross-sectional nature of data, yielding a high risk of reverse causality bias. Moreover, the measure of cognitive impairment using only an MMSE score less than 24 could lead to missing diagnoses for individuals with higher MMSE scores but impaired cognition. The measures of peripheral ARHL with both pure tone and speech audiometry could be affected by impaired cognition and the associated inability to execute the commands needed for responses in those examinations. Moreover, verbal administration of cognitive and auditory tests relies on individual peripheral and central auditory processing, and a decreased audibility can impact individual performance. For this reason, the findings of an association between age-related CAPD and cognitive frailty may be the result of the cognitive measure used (ie, MMSE) rather than a true association. In the fully adjusted models, there was a lack of adjustment for other potential confounders associated with both dementia and ARHL (ie, a family history of dementia and apolipoprotein E genotype status). In addition, age had the strongest association with ARHL and frailty among all biological factors assessed.

Conclusions

In this cross-sectional study, age-related CAPD was independently associated with cognitive frailty. To better assess the association of ARHL independently of age with the development of physical frailty and cognitive frailty, further longitudinal, epidemiologic evidence is needed, particularly in the middle-age group (40-60 years of age). A potential strategy would be an intervention trial based on the use of hearing aids and hearing restoration strategies for individuals with hearing impairment, using the development of physical and cognitive frailty phenotypes as end points.