Association between lifetime noise exposure and audiometric thresholds using the noise exposure structured interview Authors: Natela M. Shanidze, Anca Velisar, Al Lotze, Praveen Prakash, Courtney Stewart, Devin L. McCaslin Published: 2026 ABSTRACT Noise exposure is pervasive in modern society and a major contributor to hearing loss across the lifespan. This study examined how cumulative lifetime noise exposure, measured using the Noise Exposure Structured Interview (NESI), associates with age to affect audiometric thresholds. Fifty-four adults (23–82 years; 25 female) completed the NESI (occupational, recreational, and firearm noise exposure), hearing function self-reports, pure-tone audiometry, and speech-reception testing. Multivariate regression analyses evaluated independent and interactive effects of age and noise exposure on auditory outcomes. Age significantly predicted elevated thresholds across all audiometric frequencies (p < 0.001). Higher NESI scores independently associated with poorer low frequency pure tone average binaurally (p ≤ 0.02). Frequency-specific analysis showed significant noise-related contributions at 0.25–4 kHz in the worse ear and at 1.5, 2, and 6 kHz in the better ear (p < 0.03). Higher NESI scores were also associated with poorer speech reception thresholds in the worse ear (p = 0.007). These findings suggest that cumulative noise burden may accelerate age-related auditory decline. The NESI provides a practical measure of lifetime noise exposure contributing to presbycusis. Introduction In the natural world, ambient sound levels have been reported to range from 20 to 40 dB. The sounds generated by wind, birdsongs, and flowing water all fall within this range1. However, the world has grown louder and environments in urban areas can range from 55 to 70 dB with passing traffic2,3, and peak sound levels (e.g., construction work) can exceed 100 dB4. There is now compelling evidence that associates chronic noise exposure with illnesses across several body systems. These include sleep disruption and annoyance, but also stress-related illnesses (e.g., cardiovascular disease) and permanent hearing impairment5. In the U.S., noise exposure plays a critical role in the development of age-related hearing loss6,7, which affects 65.3% of adults older than 71 and 96.2% of those older than 908. Hearing loss is linked to reduced independence, cognitive impairment, multiple comorbidities9, social isolation10, and depression11. It is also recognized as a major risk factor for gait and balance deficits12 and cognitive decline13. Documenting lifetime noise exposure and integrating risk assessment into clinical care are essential steps toward preventing the broader functional and cognitive impacts of hearing loss. Although traditional clinical audiometric assessments focus primarily on hearing thresholds and speech understanding in quiet, emerging research has shown that more subtle changes in auditory physiology caused by noise exposure can result in difficulty hearing in challenging acoustic environments14. For example, research has shown that cochlear synaptopathy, a form of hearing loss caused by damage to synaptic connections between inner hair cells and auditory nerve fibers, is associated with noise exposure15. Patients with synaptopathy would be expected to have hearing difficulties in noisy environments when their audiometric thresholds would suggest they should not16,17. Accordingly, accurate documentation of a patient’s cumulative noise exposure has been increasingly advisable as an important clinical step, as lifetime noise burden may contribute to auditory dysfunction beyond what is captured by conventional threshold-based assessments alone14,18. Noise exposure has been shown extensively to initially lead to hearing loss in the 3–4 kHz range in humans (e.g19, . It has also been shown that age-related hearing loss is of particular concern in industrialized areas (e.g20, , bringing the traditional view of age-related hearing loss into question and suggesting that regular exposure to environmental noise exacerbates or even accelerates natural aging. Indeed, Wu and colleagues (2020) called into question whether age-related hearing loss is indeed noise-dominated, showing a close association between hair cell and hearing loss, with particularly basal hair cell loss being predominantly associated with noise exposure21. In animal studies of age-related hearing loss in the inner ear, research animals are kept in a controlled, low-noise environment, which likely overlooks the effects of lifetime noise exposure on inner ear (cochlear) function21. Despite this practice, it has been demonstrated that even a single, moderate noise exposure that causes a temporary hearing threshold shift can accelerate age-related hearing loss in mice22. Traditional methods of noise exposure assessment often rely on self-reported exposure histories or occupational records (NEQ23, ; LENS-Q, ONEQ). However, these approaches are hampered by variability in recall accuracy, lack of standardization, and limited attention to non-occupational exposures. Additionally, noise exposure metrics often do not extend across the lifetime or detail an individual’s use and type of hearing protection devices (HPD). These limitations fail to address the importance of comprehensively documenting a patient’s exposure risk to noise and yet there are very few standardized tools that reliably quantify lifetime noise exposure and predict associated auditory risks. One instrument that has shown promise is the Noise Exposure Structured Interview (NESI18, . The NESI was developed to address the aforementioned gaps in accurately documenting an individual’s lifetime noise exposure. The interview has been developed to incorporate best practices from existing noise exposure research and provides a quantitative exposure score. The NESI allows participants to report any activities that they identify as a risk across multiple domains: recreational, occupational, and firearm related noise. This approach facilitates a flexible individualized and comprehensive noise exposure profile. The interview is structured such that significant exposure events are documented across the individual’s life and is divided into discrete periods defined by life milestones (i.e., changes in employment). This structure is intended to facilitate recall of exposure events by associating them to salient life milestones, allowing for documenting noise exposure across the lifetime18. The NESI provides the examiner with an acoustic energy-based calculation that integrates reported exposure data across activities and time periods, yielding a cumulative noise exposure score that can be used to stratify individuals by exposure risk in research settings or to inform clinical decision-making regarding hearing conservation and protection18. The purpose of this study was to assess the association between age, lifetime noise exposure (as captured by the NESI) and hearing function (assessed through audiometric thresholds, speech perception, and self-report) of individuals across noise exposure levels and lifespan. Our primary hypothesis was that in addition to older age, greater noise exposure history would be associated with higher pure tone and speech reception thresholds. We further hypothesized that self-reported hearing difficulty/sensitivity would be greater in individuals with higher levels of lifetime noise exposure. Materials and methods Participants All research was performed in accordance with the Declaration of Helsinki and was approved by the Institutional Review Board of the Smith-Kettlewell Eye Research Institute (Protocol #SHA002). All participants provided written, informed consent to participate in the study. We recruited 54 participants (25 female, 29 male) between the ages of 23 and 82 (median: 39.5). All participants were screened for neurological and motor disorders. Individuals with a history of neurological, vestibular, or hearing pathologies were excluded. None of the participants reported prior use of ototoxic medications. Participants over 60 years of age were screened for age-related cognitive decline using the Montreal Cognitive Assessment (MoCA24, . None had scores consistent with significant cognitive impairment (score ≥ 2325, and all were thus retained in the study. Hearing was assessed using pure tone audiometry in the 0.25–8 kHz range. As a confirmatory measure, a detailed history of vestibular complaints was obtained during recruitment, and the Dizziness Handicap Inventory was administered to confirm absence of dizziness and unsteadiness in daily life26. All participants confirmed no active vestibular complaints or symptoms, ensuring the sample was not confounded by vestibular involvement. There were no significant differences between sexes across all tested frequencies (Man-Whitney U tests with Holm-Šídák correction for multiple comparisons, p > 0.1, Fig. 1A&B). Additionally, there was no difference in noise exposure history (log(NESI) scoreWelch’s t-test, t = 0.44, p = 0.66, Fig. 1C) or age (Man-Whitney U test, U = 365, p = 0.16) between sexes. Thus, for all subsequent analyses, all participants were combined. Noise exposure history Lifetime history of noise exposure was estimated using the Noise Exposure Structured Interview (NESI18, . Briefly, each participant was asked about the duration and level of noise exposure they experienced professionally, recreationally, and due to the use of firearms. Participants also had to recall what kind of ear protection (if any) was used during each activity. Each instance of a work- or leisure-related activity (e.g., commuting on the subway during college) was recorded in terms of the number of hours per day, days per week, weeks per year, and total years that the participant was engaged in that activity and the cumulative exposure was calculated using the algorithm provided by Guest et al. (2018). Each interview was conducted closely following the instructions provided in18 and modeled on the accompanying demonstration video27. Based on their responses, individuals were divided into low (log(NESI) < 1) and high (log(NESI) ≥ 1) noise exposure groups28–30. Participants were recruited widely across occupational and recreational domains to provide a broad cross-section of noise exposure profiles across all ages. To confirm that age was not the driving factor of accumulated noise exposure in our population, we performed a Spearman correlation between age and NESI score and did not find a significant relationship (ρ = 0.044, p = 0.735). This outcome was consistent with prior work examining the relationship between age and lifetime noise exposure31. Pure tone audiometry Participants underwent standard pure tone audiometry testing at 0.25, 0.5, 1, 1.5 (N = 51), 2, 3 (N = 53), 4, 6 (N = 52) and 8 kHz (N = 53) using the modified Hughson-Westlake procedure32. Hearing thresholds were determined following ISO 8253-1 guidelines33. Stimuli were presented through Etymotic ER-3 A audiometric earphones driven by an Interacoustics AC-30 audiometer. Masking was applied to the non-test ear using contralateral broadband noise when interaural threshold differences warranted it, in accordance with standard clinical masking guidelines. Normal middle ear function was confirmed using bone conduction testing at 0.25, 0.5, 1, 2, and 4 kHz, with air-bone gaps not exceeding 10 dB, and tympanometry using a 226 Hz probe tone (Amplivox Otowave 102), with Type A tympanograms confirming normal middle ear pressure and compliance. For the purposes of analysis and classification, the worse and better hearing ears were determined for each participant using the average threshold across all tested frequencies. Across all participants, the average difference between the ears was 4.6 dB, with five participants having an interaural threshold difference greater than 10 dB (11, 13, 14, 15, 39 dB HL). The worse ear was analyzed as it is expected to reflect the cumulative burden of noise exposure most sensitively, while the better ear provides a within-subject reference less confounded by asymmetric exposure history. Fig. 1Population auditory thresholds for worse (A) and better (B) hearing ears and log(NESI) scores (C) for males (blue) and females (gray). Population auditory thresholds for worse (A) and better (B) hearing ears and log(NESI) scores (C) for males (blue) and females (gray). Hearing function self-report All participants completed previously developed questionnaires34 examining difficulty in quiet (1 item, “You are in a group of about five people, sitting round a table. It is an otherwise quiet place. You can see everyone else in the group. Can you follow the conversation?”) and noisy environments (4 items, e.g., “You are having a conversation with one person in a room where there are many other people talking. Can you follow what the person you are talking to is saying?”), ability to localize a sound source (3 items, e.g., “You are in an unfamiliar house. It is quiet. You hear a door slam. Can you tell right away where that sound came from?”; all three on a 0–10 Likert scale), as well as loudness and annoyance of sounds (12 items, e.g., “barking dog next to you”, “someone chewing food or gum”), and hyperacusis (10 items, e.g., “because sounds are too loud you avoid shopping”, “you feel anxious”; all three on a 0-100 scale. Full questionnaires are provided as supplementary materials in34. Questions in this questionnaire are analogous to a subset of the questions included in the Speech, Spatial and Qualities of Hearing Scale (SSQ)35. Speech reception thresholds Speech reception thresholds (SRT) were obtained for a subset of 33 participants for each ear using a standard recording of the C.I.D. W-1 list of spondaic words (Auditec, Inc. item 102SI). The participants were instructed to repeat the words verbally. Words were presented in groups of three, starting at 70 dB SL with a 10 dB reduction in level for each three-word sequence until the participant was unable to repeat a minimum of two words in a group correctly. The level was then increased in 5 dB steps until a minimum of two correct responses per word group were obtained. Additional stimulus presentations were then made, reducing the level in 10 dB steps when at least two correct responses were obtained and raising it in 5 dB steps when fewer than two correct responses were obtained, until the same response threshold was obtained a minimum of three out of five times. Masking was applied to the non-test ear when necessary, following standard clinical masking criteria. Participants tested on the SRT paradigm had similar accumulated noise exposure histories and hearing function as those who did not perform the SRT (log(NESI): Mann-Whitney U = 313, p = 0.56; worse and better ear hearing thresholds: Mann-Whitney U tests with Holm-Šídák correction for multiple frequency comparisons, all p\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\ge\:$$\end{document}0.17 for both ears). Analysis & regression modeling Across all analyses, normality was assessed using the Anderson-Darling test. Statistical analyses were performed using Prism 10 (GraphPad Software LLC) and custom MATLAB scripts. Normality testing showed that auditory threshold data did not follow a Gaussian distribution, and further examination indicated that the Gamma distribution provided the best fit to the data. To accommodate the constraints of this distribution for generalized linear modeling (see below), all threshold data were uniformly shifted by 20 dB for pure tone audiometry and 10 dB for SRT to ensure that all values were strictly positive. The offset was reversed for subsequent outcome illustration and interpretation. One participant had significantly higher interaural asymmetry than the rest of the group (39 dB). Although this participant met all inclusion criteria, we wanted to ensure that they did not have an outsized influence on model outcomes. We compared the models with this participant included and removed and examined if their removal led to significant changes in the F values, changes in the sign of the coefficients, and if p become no longer significant. The models were robust, with all three criteria not changing whether that participant was or was not included. Based on these outcomes, we retained the participant in our analysis. Model selection To determine the relationship between noise exposure and hearing thresholds, we constructed a series of generalized linear models (GLMs) for the low and high frequency pure tone averages for each ear (LFPTA: [0.5, 1, 1.5, 2] kHz and HFPTA:3,4,6,8 kHz, respectively), each tested frequency, and SRT. For each of these dependent variables, we tested separate hypotheses regarding their relationship with the predictor variables (age, lifetime noise exposure). Our primary hypotheses were that higher age and log(NESI) would be associated with elevated auditory thresholds for LFPTA and HFPTA. Based on the model outcomes of these two comparisons, we then performed secondary analyses for each independent frequency band tested to determine whether there were specific frequencies that were driving outcomes in the primary analysis. For model selection, age was always included in the model (treated as a mandatory factor), while we tested whether the addition of log(NESI) and their interaction term meaningfully improved the model (candidate variables). For each dependent variable, models were constructed using the identity link function with normal distribution for ordinary linear models (OLS) and reciprocal (1/µ) and log (ln(µ)) link functions for generalized linear models (GLM) with gamma distribution. Models were compared using the Akaike Information Criterion (AIC). The model with the lowest AICc (AIC corrected for smaller samples) was chosen and the corresponding link function is indicated in Table 1. Interpretation of predictor coefficients for GLM with the two selected link functions: ln(µ) – one unit increase in the value of the predictor determines the multiplication of the estimated mean of the response variable by eβ. The percentage change of the mean of the response variable will be given by (eβ-1)x100. For example, for a β = 0.025, an increase of one year in age would lead to an increase of the average HFPTA in the worse ear by 2.5%. A negative coefficient will lead to an analogous decrease.1/µ - if β > 0, an increase in the value of the predictor determines a decrease in the average of the response variable, while if β < 0, an increase in the value of the predictor determines an increase in the average of the response variable. For example, for a negative β, an increase in age would lead to an increase in average LFPTA. ln(µ) – one unit increase in the value of the predictor determines the multiplication of the estimated mean of the response variable by eβ. The percentage change of the mean of the response variable will be given by (eβ-1)x100. For example, for a β = 0.025, an increase of one year in age would lead to an increase of the average HFPTA in the worse ear by 2.5%. A negative coefficient will lead to an analogous decrease. ln(µ) – one unit increase in the value of the predictor determines the multiplication of the estimated mean of the response variable by eβ. The percentage change of the mean of the response variable will be given by (eβ-1)x100. For example, for a β = 0.025, an increase of one year in age would lead to an increase of the average HFPTA in the worse ear by 2.5%. A negative coefficient will lead to an analogous decrease. 1/µ - if β > 0, an increase in the value of the predictor determines a decrease in the average of the response variable, while if β < 0, an increase in the value of the predictor determines an increase in the average of the response variable. For example, for a negative β, an increase in age would lead to an increase in average LFPTA. 1/µ - if β > 0, an increase in the value of the predictor determines a decrease in the average of the response variable, while if β < 0, an increase in the value of the predictor determines an increase in the average of the response variable. For example, for a negative β, an increase in age would lead to an increase in average LFPTA. For parameter selection, the simplest model that included only age and log-transformed NESI scores as predictor variables was subsequently compared against a model that included an age-by-NESI interaction term. The interaction term did not improve any of the models. Results Auditory thresholds as a function of self-reported noise exposure First, we examined the relationship between auditory thresholds, age, and cumulative noise exposure (log-transformed NESI scores) for all participants for the low (LFPTA) and high frequency pure tone averages (HFPTA). Age was a strong predictor for both LFPTA and HFPTA for both better and worse ears (3rd and 8th columns, Table 1, p < 0.0001). For the LFPTA, log(NESI) was also a significant predictor for both ears (4th and 9th columns, Table 1, p < 0.05). Given the above outcomes in both the low and high frequency averages, we repeated the analysis across all tested frequencies individually (Table 1). Age was a strong predictor of auditory threshold for all tested frequencies, in both worse and better-hearing ears (3rd and 8th columns, Table 1). Noise exposure became a significant predictor at frequencies as low as 0.25 kHz in the worse ear (Column 4, Table 1) and 1.5 kHz in the better ear (Column 8, Table 1). Across all frequencies, a higher NESI score and greater age were both associated with worse thresholds (see Interpretation of Predictor Coefficients in Sect. 2.6.1 of Methods). Effects of cumulative noise exposure on age-related increases in auditory thresholds Figure 2 shows the relationships between LFPTA (Fig. 2A, C)/HFPTA (Fig. 2B, D), age, and cumulative lifetime noise exposure (log(NESI)) estimated by the models described in Table 1. For both worse (Fig. 2A, B) and better ears (Fig. 2C, D) there is a clear increase in average thresholds with age (right horizontal axis). This increase is greater in individuals with greater noise exposure (left horizontal axis). The model surface shown in Fig. 2 allows one to estimate a PTA threshold for any given combination of age and log(NESI) score. While thresholds are around 0 dB in all four panes for young individuals (~ 20 years) with limited noise exposure (~ 0), thresholds can be in the 60–70 s dB for older adults (~ 80 years) with high levels of noise exposure (~ 2). We provide threshold estimates for this latter category in the 6th and 11th columns of Table 1 and have marked them of low and high frequency pure tone averages in Fig. 2 with a black cross. These values (in the yellow – high – portion of the surface) are significantly higher than those for older adults with low/no levels of noise exposure, particularly for the LFPTA (blue in Fig. 2A & C). Table 1GLM parameters of auditory threshold as a function of age and lifetime noise exposure.Freq. Band(kHz)Worse EarBetter EarLINKAgelog(NESI)pseudoR2 (%) & DFPredicted Threshold:Age = 80log(NESI) = 2LINKAgelog(NESI)pseudoR2 (%) & DFPredicted Threshold:Age = 80log(NESI) = 2 LFPTA [0.5, 1, 1.5, 2]1/β = -0.0004p < 0.0001f2 = 1.15β = -0.0037p = 0.001f2 = 0.3564.005151.771/β = -0.0005p < 0.0001f2 = 1.07β = -0.0028p = 0.021f2 = 0.1262.065135.04 HFPTA [3, 4 6, 8]ln()β = 0.0247p < 0.0001f2 = 4.04β = 0.0555p = 0.189f2 = 0.1482.025174.25ln()β = 0.0243p < 0.0001f2 = 4.11β = 0.0672p = 0.096f2 = 0.1782.755163.30 0.25 1/β = -0.0004p < 0.0001f2 = 1.24β = -0.0022p = 0.048f2 = 0.1255.115138.511/β = -0.0004p < 0.0001f2 = 1.05β = -0.0017p = 0.202f2 = 0.0448.505130.00 0.5 1/β = -0.0004p < 0.0001f2 = 0.96β = -0.0017p = 0.159f2 = 0.0450.965138.431/β =-0.0004p < 0.0001f2 = 0.97β = -0.0017p = 0.217f2 = 0.0348.225127.78 1 1/β = -0.0004p < 0.0001f2 = 0.80β = -0.0044p = 0.003f2 = 0.2857.315150.551/β = -0.0004p < 0.0001f2 = 0.68β = -0.0024p = 0.101f2 = 0.0549.095128.22 1.5 1/β = -0.0004p < 0.0001f2 = 0.94β = -0.0043p = 0.002f2 = 0.3859.204853.441/β = -0.0005p < 0.0001f2 = 0.73β = -0.0035p = 0.031f2 = 0.1454.444836.96 2 1/β = -0.0004p < 0.0001f2 = 1.13β = -0.0042p = 0.001f2 = 0.5057.175167.471/β = -0.0005p < 0.0001f2 = 1.10β = -0.0033p = 0.008f2 = 0.1764.105149.06 3 1/β = -0.0005p < 0.0001f2 = 2.31β = -0.0021p = 0.027f2 = 0.3971.155088.94ln()β = 0.0226p < 0.0001f2 = 1.53β = -0.0090p = 0.870f2 = -0.0166.375048.66 4 1/β = -0.0005p < 0.0001f2 = 2.28β = -0.0026p = 0.018f2 = 0.4067.0251100.53ln()β = 0.0234p < 0.0001f2 = 2.28β = 0.0922p = 0.107f2 = 0.1168.155157.85 6 ln()β = 0.0241p < 0.0001f2 = 4.44β = 0.0627p = 0.136f2 = 0.1681.444977.48ln()β = 0.0238p < 0.0001f2 = 3.72β = 0.0926p = 0.026f2 = 0.2880.474971.72 8 ln()β = 0.0328p < 0.0001f2 = 2.28β = 0.0437p = 0.474f2 = 0.0379.6951100.36ln()β = 0.0284p < 0.0001f2 = 4.90β = 0.0853p = 0.110f2 = 0.1979.025176.92 SRT 1/β = -0.0008p < 0.0001f2 = 0.83β = -0.0067p = 0.007f2 = 0.3545.33063.93ln()β = 0.0160p < 0.0001f2 = 0.64β = 0.1006p = 0.132f2 = 0.0635.53020.27**All model fits are significant at p < 0.0001. Cohen’s f2 indicates effect size for each β estimate. Table 1GLM parameters of auditory threshold as a function of age and lifetime noise exposure.Freq. Band(kHz)Worse EarBetter EarLINKAgelog(NESI)pseudoR2 (%) & DFPredicted Threshold:Age = 80log(NESI) = 2LINKAgelog(NESI)pseudoR2 (%) & DFPredicted Threshold:Age = 80log(NESI) = 2 LFPTA [0.5, 1, 1.5, 2]1/β = -0.0004p < 0.0001f2 = 1.15β = -0.0037p = 0.001f2 = 0.3564.005151.771/β = -0.0005p < 0.0001f2 = 1.07β = -0.0028p = 0.021f2 = 0.1262.065135.04 HFPTA [3, 4 6, 8]ln()β = 0.0247p < 0.0001f2 = 4.04β = 0.0555p = 0.189f2 = 0.1482.025174.25ln()β = 0.0243p < 0.0001f2 = 4.11β = 0.0672p = 0.096f2 = 0.1782.755163.30 0.25 1/β = -0.0004p < 0.0001f2 = 1.24β = -0.0022p = 0.048f2 = 0.1255.115138.511/β = -0.0004p < 0.0001f2 = 1.05β = -0.0017p = 0.202f2 = 0.0448.505130.00 0.5 1/β = -0.0004p < 0.0001f2 = 0.96β = -0.0017p = 0.159f2 = 0.0450.965138.431/β =-0.0004p < 0.0001f2 = 0.97β = -0.0017p = 0.217f2 = 0.0348.225127.78 1 1/β = -0.0004p < 0.0001f2 = 0.80β = -0.0044p = 0.003f2 = 0.2857.315150.551/β = -0.0004p < 0.0001f2 = 0.68β = -0.0024p = 0.101f2 = 0.0549.095128.22 1.5 1/β = -0.0004p < 0.0001f2 = 0.94β = -0.0043p = 0.002f2 = 0.3859.204853.441/β = -0.0005p < 0.0001f2 = 0.73β = -0.0035p = 0.031f2 = 0.1454.444836.96 2 1/β = -0.0004p < 0.0001f2 = 1.13β = -0.0042p = 0.001f2 = 0.5057.175167.471/β = -0.0005p < 0.0001f2 = 1.10β = -0.0033p = 0.008f2 = 0.1764.105149.06 3 1/β = -0.0005p < 0.0001f2 = 2.31β = -0.0021p = 0.027f2 = 0.3971.155088.94ln()β = 0.0226p < 0.0001f2 = 1.53β = -0.0090p = 0.870f2 = -0.0166.375048.66 4 1/β = -0.0005p < 0.0001f2 = 2.28β = -0.0026p = 0.018f2 = 0.4067.0251100.53ln()β = 0.0234p < 0.0001f2 = 2.28β = 0.0922p = 0.107f2 = 0.1168.155157.85 6 ln()β = 0.0241p < 0.0001f2 = 4.44β = 0.0627p = 0.136f2 = 0.1681.444977.48ln()β = 0.0238p < 0.0001f2 = 3.72β = 0.0926p = 0.026f2 = 0.2880.474971.72 8 ln()β = 0.0328p < 0.0001f2 = 2.28β = 0.0437p = 0.474f2 = 0.0379.6951100.36ln()β = 0.0284p < 0.0001f2 = 4.90β = 0.0853p = 0.110f2 = 0.1979.025176.92 SRT 1/β = -0.0008p < 0.0001f2 = 0.83β = -0.0067p = 0.007f2 = 0.3545.33063.93ln()β = 0.0160p < 0.0001f2 = 0.64β = 0.1006p = 0.132f2 = 0.0635.53020.27**All model fits are significant at p < 0.0001. Cohen’s f2 indicates effect size for each β estimate. GLM parameters of auditory threshold as a function of age and lifetime noise exposure. LFPTA [0.5, 1, 1.5, 2] β = -0.0004 p < 0.0001 f2 = 1.15 β = -0.0037 p = 0.001 f2 = 0.35 64.00 51 β = -0.0005 p < 0.0001 f2 = 1.07 β = -0.0028 p = 0.021 f2 = 0.12 62.06 51 HFPTA [3, 4 6, 8] β = 0.0247 p < 0.0001 f2 = 4.04 β = 0.0555 p = 0.189 f2 = 0.14 82.02 51 β = 0.0243 p < 0.0001 f2 = 4.11 β = 0.0672 p = 0.096 f2 = 0.17 82.75 51 β = -0.0004 p < 0.0001 f2 = 1.24 β = -0.0022 p = 0.048 f2 = 0.12 55.11 51 β = -0.0004 p < 0.0001 f2 = 1.05 β = -0.0017 p = 0.202 f2 = 0.04 48.50 51 β = -0.0004 p < 0.0001 f2 = 0.96 β = -0.0017 p = 0.159 f2 = 0.04 50.96 51 β =-0.0004 p < 0.0001 f2 = 0.97 β = -0.0017 p = 0.217 f2 = 0.03 48.22 51 β = -0.0004 p < 0.0001 f2 = 0.80 β = -0.0044 p = 0.003 f2 = 0.28 57.31 51 β = -0.0004 p < 0.0001 f2 = 0.68 β = -0.0024 p = 0.101 f2 = 0.05 49.09 51 β = -0.0004 p < 0.0001 f2 = 0.94 β = -0.0043 p = 0.002 f2 = 0.38 59.20 48 β = -0.0005 p < 0.0001 f2 = 0.73 β = -0.0035 p = 0.031 f2 = 0.14 54.44 48 β = -0.0004 p < 0.0001 f2 = 1.13 β = -0.0042 p = 0.001 f2 = 0.50 57.17 51 β = -0.0005 p < 0.0001 f2 = 1.10 β = -0.0033 p = 0.008 f2 = 0.17 64.10 51 β = -0.0005 p < 0.0001 f2 = 2.31 β = -0.0021 p = 0.027 f2 = 0.39 71.15 50 β = 0.0226 p < 0.0001 f2 = 1.53 β = -0.0090 p = 0.870 f2 = -0.01 66.37 50 β = -0.0005 p < 0.0001 f2 = 2.28 β = -0.0026 p = 0.018 f2 = 0.40 67.02 51 β = 0.0234 p < 0.0001 f2 = 2.28 β = 0.0922 p = 0.107 f2 = 0.11 68.15 51 β = 0.0241 p < 0.0001 f2 = 4.44 β = 0.0627 p = 0.136 f2 = 0.16 81.44 49 β = 0.0238 p < 0.0001 f2 = 3.72 β = 0.0926 p = 0.026 f2 = 0.28 80.47 49 β = 0.0328 p < 0.0001 f2 = 2.28 β = 0.0437 p = 0.474 f2 = 0.03 79.69 51 β = 0.0284 p < 0.0001 f2 = 4.90 β = 0.0853 p = 0.110 f2 = 0.19 79.02 51 β = -0.0008 p < 0.0001 f2 = 0.83 β = -0.0067 p = 0.007 f2 = 0.35 45.3 30 β = 0.0160 p < 0.0001 f2 = 0.64 β = 0.1006 p = 0.132 f2 = 0.06 35.5 30 **All model fits are significant at p < 0.0001. Cohen’s f2 indicates effect size for each β estimate. Fig. 2Observed values (dots) and model fit (mesh grid) of low (A, C) and high (B, D) pure tone averages as a function of log(NESI) (left horizontal axis) and age (right horizontal axis) for worse (A, B) and better (C, D) ears. Red dots: values above the model estimate, purple dots: values below the model estimate, blue lines: model values at low (-2, 0) and high (2) log(NESI) values for younger (30 y.o.) and older (80 y.o.) adults. Black cross: PTA threshold estimate at Age = 80, log(NESI) = 2 (provided in columns 6 and 11 of Table 1). Observed values (dots) and model fit (mesh grid) of low (A, C) and high (B, D) pure tone averages as a function of log(NESI) (left horizontal axis) and age (right horizontal axis) for worse (A, B) and better (C, D) ears. Red dots: values above the model estimate, purple dots: values below the model estimate, blue lines: model values at low (-2, 0) and high (2) log(NESI) values for younger (30 y.o.) and older (80 y.o.) adults. Black cross: PTA threshold estimate at Age = 80, log(NESI) = 2 (provided in columns 6 and 11 of Table 1). Association of hearing self-reports with noise exposure history We examined differences in self-reported ability to listen in quiet and loud environments, and localize sounds, as well as annoyance, loudness, and avoidance (hyperacusis) of sounds (Fig. 334, in participants with low (logNESI < 1) and high (logNESI ≥ 1) lifetime noise exposure. We found a significant difference between groups only for the ability to listen in quiet environments (Mann-Whitney U t-test, with Holm-Šídák correction for multiple comparisons, adjusted p = 0.02). Fig. 3Comparison of self-reported sound ability (A) and sensitivity, and hyperacusis (B) between individuals who had low (purple) or high (turquoise) levels of noise exposure. Comparison of self-reported sound ability (A) and sensitivity, and hyperacusis (B) between individuals who had low (purple) or high (turquoise) levels of noise exposure. Speech reception threshold Speech reception threshold (SRT) was examined as an additional outcome measure given its clinical relevance as an index of functional hearing ability. While SRT typically correlates with the low-frequency pure tone average in clinical practice, accurate speech intelligibility demands finer cochlear coding of spectral and temporal features of speech signals beyond simple threshold detection. Therefore, SRT was included to determine whether cumulative noise exposure predicts functional speech audibility over and above what pure tone thresholds alone would capture, given that cochlear damage from noise may affect the fine structure coding necessary for speech intelligibility even when threshold elevation is modest. For a subset of 33 participants, an analogous analysis to that for pure tone audiometry was performed (Table 1, bottom row). For the worse ear, we found that akin to the LFPTA both age and log(NESI) were significant predictors. For the better ear, only age was a significant predictor, while log(NESI) did not have a significant contribution to the model fit (Table 1, bottom row). Once again, the simplest model was preferred based on the model AICcs. Discussion Summary of the study findings The primary purpose of this study was to investigate the association between the age of participants, lifetime noise exposure using the NESI, and audiometric thresholds. Our findings show that age and cumulative noise exposure significantly influenced auditory thresholds, at frequencies as low as 0.25 kHz, as well as at higher frequencies (3, 4, and 6 kHz, see Table 1). Individuals with a higher lifetime noise exposure (i.e., NESI score) showed a steeper association between age and hearing threshold elevation, suggesting a greater age-related threshold shift in those with higher cumulative noise burden (see Fig. 2). This phenomenon was particularly evident in the low-frequency pure tone average (Fig. 2A and C), despite the typical noise notch including 3 and 4 kHz frequencies19. The driving factor for this outcome is likely presbycusis being more severe in the 4–8 kHz frequency range36,37 as seen in the higher threshold versus age relationship for the lowest noise exposures in Fig. 2B and D. Our data demonstrate a potential noise notch that is wider than the current literature suggests19. In the better ear, significant contribution of noise exposure at 3 and 6 kHz is observed, which agrees with the established literature. In the worse ear, the interaction is observed as low as 0.25 kHz. Given that the apical cochlear region corresponding to 0.25 kHz is generally considered less susceptible to conventional air-conducted noise injury, the precise mechanisms underlying this observation warrant further investigation. The extension of NESI-related effects into lower frequencies is notable and suggests that cumulative environmental noise exposure may contribute to a broader pattern of cochlear vulnerability than predicted by the classical 3–4 kHz noise notch. Although the apical cochlear regions are generally considered less susceptible to conventional noise injury, the association between cumulative noise exposure and 0.25 kHz thresholds observed here may reflect an interaction between age-related cochlear changes, particularly metabolic degeneration of the stria vascularis, and cumulative noise burden, rather than direct noise-induced apical damage. This interpretation is consistent with the cross-sectional nature of the data, which captures the combined influence of these factors rather than isolating either independently. Deviations from the canonical noise-notch configuration have also been described in population-based and experimental studies of age-related and noise-related auditory decline37,38. The findings also suggest that the worse-hearing ear may serve as an early sentinel marker of noise-induced cochlear vulnerability. While cochlear synaptopathy may also contribute to the cumulative noise burden, it predominantly manifests as suprathreshold processing deficits rather than threshold elevation and thus cannot be directly inferred from the present audiometric data. Experimental evidence indicates that noise exposure can induce cochlear synaptopathy even in cases where conventional audiometric thresholds remain within normal limits15,17. These findings suggest that lifetime noise burden may accelerate neural aspects of age-related auditory decline. Such a mechanism could account for the additional variance in threshold outcomes explained by NESI scores beyond the effects of age alone in the present cohort. However, the relationship between age, noise exposure, and cochlear synaptopathy remains complex; Prendergast et al. (2019) demonstrated that while both age and lifetime noise exposure predicted speech-in-noise performance, the overall pattern of findings was inconsistent with predictions from the low-spontaneous rate synaptopathy hypothesis, underscoring the difficulty of isolating noise-induced neural changes from age-related auditory decline39. Use of NESI in prior work The current study’s findings are in line with and add to previous research showing that degree of lifetime noise exposure significantly exacerbates age-related hearing loss, previously demonstrated for high frequencies17,40. However, there is a dearth of research as it relates to more clearly understanding the relationship between amount of accumulated noise exposure as measured with the NESI and sensorineural hearing loss. Despite the important information regarding noise exposure that is captured with the NESI, only a few studies have employed the tool to date (e.g30,41,42., ). In a longitudinal study of early career musicians, NESI-based estimates of lifetime noise exposure were associated with reductions in distortion product otoacoustic emissions (DPOAEs) and an increase in self-reported tinnitus symptoms over time, despite audiometric thresholds remaining within normal limits43. The findings from this investigation suggest that the NESI can reveal important links between cumulative noise exposure and subtle physiological changes in the auditory system that conventional audiometry may miss. In addition, an ongoing feasibility study for the Nottingham Hearing BioResource is employing the NESI to estimate lifetime noise exposure in a large general population sample, with the goal of linking detailed noise histories to audiological, cognitive, and genetic data over time44. Couth et al. (2024) followed early-career musicians longitudinally using the NESI and found that higher cumulative NESI scores were associated with reduced DPOAE amplitudes, despite stable audiometric thresholds, suggesting the NESI can capture subtle physiological changes that conventional audiometry may miss43. Others are developing protocols that incorporate the NESI (e.g44, into large-scale studies, highlighting interest in the calculation of noise exposure history as an important component of audiological examination and longitudinal hearing research. Collectively, these findings suggest that the NESI shows promise as a research tool for quantifying lifetime noise burden and linking it to measurable auditory changes. While direct evidence for its impact on clinical decision-making remains limited, the present findings add to a growing body of work supporting its potential utility in audiological evaluation, particularly for identifying individuals at greater risk of hearing loss than expected for their age. The clinical utility of the NESI is that this tool can provide valuable and unique insight into an individual’s history of noise exposure. By identifying individuals who may be at risk earlier, clinicians can recommend proactive interventions that can mitigate or prevent noise-induced hearing loss. Surprisingly, this tool is largely limited to the lab18,45–47 as compared to the clinic43 when contrasted to other popular self-report measures of hearing handicap, disability, and impairment. From a clinical standpoint, the present findings highlight the potential value of incorporating structured lifetime noise exposure assessment into routine audiologic evaluation, particularly for adults who present with subtle threshold changes that might otherwise be attributed to aging. Because the NESI can be administered efficiently and captures exposure across occupational, recreational, and firearm-related domains, its broader integration into clinical workflows may enhance early risk stratification and support more individualized hearing conservation counseling18,43,44. Areas for future study The findings of the current study have identified several areas for future research. First, broadening the study’s recruitment to expand our demographics is necessary. Our Age × NESI analyses illustrated that noise is associated with age-related increases in auditory thresholds. For example, in Fig. 2, the steepest increases in auditory thresholds with age are seen for log(NESI) > 1 regions of the modeled surface, particularly for the LFPTA, where the steepest (yellow) regions are only present with very high lifetime noise exposure (Fig. 2A&C). These patterns are aligned with reports showing that this is the frequency range in which early noise-related pathology is most likely to occur19. To validate observations from this study, we will need to work towards specifically recruiting vulnerable populations like musicians and construction workers. These populations have been identified as high-risk in recent phenotyping initiatives such as the Nottingham Hearing BioResource44. The findings from our data also confirm that the NESI can capture meaningful variance in hearing outcomes. The finding of an Age - NESI interaction was significant for low and high frequency pure tone averages, and participants reporting severe lifetime exposure exhibited thresholds 10–15 dB poorer by mid-life than those participants reporting mild lifetime exposure. Developing a method to integrate the NESI into routine audiometric assessment may provide the clinician with the chance to engage the patient on whether hearing-protection devices are recommended and also provide an opportunity to track, at least annually, the accumulation of noise exposure over time. This may lead to the development of noise dose-response data that will inform occupational safety, environmental health, and improved safety measures in recreational activities to reduce noise exposure risk. Future work should be undertaken to quantify the perceived barriers to adopting the NESI as a routine component of the audiologic test battery. Audiologic care spans tertiary medical centers, community hospitals, and private practices, and resistance to workflow change may vary across these settings and among audiologists, physicians, and advanced practice providers (APPs). Undertaking a survey of clinical providers on how to integrate the NESI into the clinical workflow should be a core effort. Finally, how the lifetime cumulative noise exposure associates with speech reception thresholds warrants further investigation. In the worse ear, log(NESI) was a significant predictor of SRT. This finding is consistent with previous work showing no significant association between lifetime noise exposure, as measured by the NESI, and speech-in-noise performance or self-reported hearing difficulties18,31. It is possible that other measures of speech-in-noise may correlate better with self-report measures including Difficulty Listening in Noisy Environments. The SRT measures the minimum hearing level at which subjects can identify speech, a different test, the Words-in-Noise (WIN) test assesses speech intelligibility in noisy environments. It is possible that subjects that are noise exposed, older, or both; can hear words in noise but have a greater level of difficulty processing this information. Future work should incorporate other metrics such as WIN to evaluate ability to comprehend speech in noise as well as ability to hear. Limitations There are several limitations to the investigation. The retrospective self-reporting for noise exposure history, despite NESI’s structured format, may introduce recall bias. Cross-sectional data collection also limits, to some degree, the causal inferences. These limitations emphasize the importance of pursuing a longitudinal study of hearing function that incorporates self-report measures, quantitative auditory, behavioral, and electrophysiology studies with a cumulative history of noise exposure (NESI). Conclusions The findings from our study illustrate the significant impact of both age and cumulative noise exposure on auditory thresholds and endorse the NESI’s utility as a valuable investigative tool. Furthermore, our findings make a strong argument for broader adoption of the NESI into clinical practice and introduce a number of questions regarding how the instrument can be used to mitigate noise-induced hearing loss, inform clinicians regarding the need for hearing protection, and promote auditory health. From a broader population health perspective, the present findings add to a growing body of evidence indicating that age-related hearing loss should not be regarded solely as an inevitable consequence of biological aging. Rather, cumulative environmental exposures, particularly modifiable factors such as noise are likely to influence the trajectory of auditory decline across the lifespan. Integrating lifetime noise exposure metrics into epidemiologic and clinical frameworks may therefore be important for informing more effective prevention strategies in increasingly urbanized and acoustically complex environments7,8,40.