Introduction

Tinnitus is the perception of sound in the absence of an external stimulus.1 Persistent tinnitus can be disabling, adversely impacting sleep, work, daily function, and quality of life.2 An estimated 50 million Americans suffer from tinnitus, among whom 3 million are severely disabled by it.3 There are few long-term studies that have assessed the incidence of tinnitus.4, 5 In most individuals with tinnitus, the etiology of their tinnitus is often unknown and the effectiveness of proposed treatments is uncertain.6 Therefore, identifying potentially modifiable risk factors could aid in prevention and potentially inform targeted treatments for this challenging condition.

Over-the-counter analgesics, including aspirin, non-salicylate non-steroidal anti-inflammatory drugs (NSAIDs), and acetaminophen, are widely used in the USA and globally.7 In a given week, at least one of these medications is taken by approximately 20% of the US adult population.8, 9 Selective cyclo-oxygenase 2 (COX-2) inhibitors are prescription NSAIDs used to achieve analgesic and anti-inflammatory properties similar to non-prescription NSAIDS but with less GI toxicity.10, 11

Analgesics may influence auditory function and several mechanisms may underlie the potential ototoxicity of these medications, including alterations in cochlear blood flow and function, response to inflammation, susceptibility to oxidative damage, and neurotransmission along central auditory pathways.12,13,14,15,16 Frequent analgesic use has been associated with higher risk of hearing loss in men and women.17,18,19 While the association of high doses of aspirin and reversible tinnitus is well-recognized,20 the associations of typical doses of analgesic medications and persistent tinnitus are not known. Therefore, we longitudinally examined the independent associations of analgesic use, specifically low-dose and moderate-dose aspirin, NSAIDs, acetaminophen, and the risk of incident persistent tinnitus in a well-characterized ongoing prospective cohort of 69,455 women in the Nurses’ Health Study II (NHSII).

Methods

Study Population

The Conservation of Hearing Study (CHEARS) examines risk factors for hearing loss and tinnitus among participants in several large ongoing prospective cohort studies, including NHSII. NHSII was established in 1989 with the enrollment of 116,430 female registered nurses aged 25 to 42 years. Participants completed questionnaires at baseline and every 2 years regarding a wide range of demographic, health, diet, and lifestyle factors, including detailed information on medication use (questionnaires can be accessed at: https://www.nurseshealthstudy.org/participants/questionnaires). The cohort has lost only 5% of person-time during follow-up.21 Participants with tinnitus at the 1995 study baseline (the first year that frequency of analgesic use was assessed) were excluded, allowing us to examine the longitudinal association of analgesic use and subsequent development of tinnitus. We also excluded women who reported cancer (other than non-melanoma skin cancer) due to potential exposure to ototoxic chemotherapy. In our primary analysis, 69,455 women were included in the analytic sample. The study protocol was approved by the Institutional Review Board of Brigham and Women’s Hospital.

Assessment of Analgesic Intake

On the 1995 biennial questionnaire and every 2 years thereafter, women were asked about regular use over the previous 2 years of aspirin, acetaminophen, and NSAIDs, along with specific examples for each type of analgesic. In a previous study of analgesic use in NHSII, ibuprofen accounted for 80% of NSAID use in this cohort.22 Information on use in days/week (1, 2–3, 4–5, 6+ days/week) and the number of tablets/week (1–15+ tablets/week) was obtained. Beginning in 2001, information on use in days/week of low-dose aspirin (≤100 mg), moderate-dose aspirin (≥325 mg), or aspirin-containing products, and on regular use (2+ days/week) of COX-2 inhibitors (e.g., celecoxib) was obtained. Although subsequent questionnaires obtained information on use in days/week of COX-2 inhibitors, we categorized use as <2 days/week(referent) and 2+ days/week due to the limited numbers of exposed cases. Analgesic use assessed in this manner has been shown to be associated with a number of important health outcomes, including hypertension,22 cancer,23 Parkinson’s disease,24 and hearing loss.17, 18

Ascertainment of Tinnitus

Information on tinnitus was collected on the 2009, 2013, and 2017 biennial questionnaires. Participants were asked, “In the past 12 months, have you had ringing, roaring, or buzzing in your ears or head?” Information was also collected on how often the symptoms occur (ranging from never to every day), how long symptoms last (ranging from “a few seconds” to “all the time”), and when the symptoms first began. Consistent with methods used previously in this and other cohorts,25,26,27 we defined persistent tinnitus as tinnitus occurring several days/week or more. In additional analyses, we examined alternative definitions of tinnitus, including tinnitus lasting 5 min or longer and occurring every day.

Assessment of Covariates

In multivariable-adjusted analyses, we included factors potentially associated with analgesic use and with tinnitus, using updated information. Factors included were age, race, body mass index, waist circumference, physical activity, smoking, hypertension, diabetes mellitus, rheumatoid arthritis, systemic lupus erythematosus, other arthritis, migraine headache, caffeine intake, dietary intake (using the Alternate Healthy Eating Index (AHEI-2010), a validated measure of diet quality based on the USDA Dietary Guidelines for Americans and dietary factors related to chronic disease),28 alcohol intake, use of thiazide diuretics and furosemide, depression (clinician-diagnosed or use of antidepressant medications), anxiety (using the phobic anxiety scale of the Crown-Crisp Experiential Index (CCI),29 an 8-item scale focused on symptoms relevant to fear and phobia disorders, such as panic disorder and agoraphobia, that also incorporates worry-related items, administered in 1993 and 2005), and self-reported hearing status. The CCI has been used successfully to examine a number of psychiatric, psychological, and social determinants of health in this and similar cohorts.30 Covariate information on many of these factors has been validated.21, 31,32,33,34 Information on analgesic use and time-varying covariates was updated at the beginning of each time period based on information obtained from the most recent questionnaire cycle.

Statistical Methods

We obtained analgesic data prospectively, before onset of tinnitus. To account for changes in participants’ patterns of use over several years, we updated analgesic data biennially and used the most recent exposure and risk factor data for each 2-year time period. Person-time of follow-up was calculated from the date of return of the baseline questionnaire for that analysis until the date of onset of tinnitus, death, or end of follow-up, whichever came first. Participants with cancers other than non-melanoma skin cancer were censored when reported during follow-up. Women without regular use of the individual analgesics served as the referent group. Person-months of follow-up were allocated according to exposure status at the start of each follow-up period. If a participant reported tinnitus that occurred less than once/week or once/week, that participant was skipped for that time period but could re-enter the analysis if tinnitus occurring several days/week or more was reported in a subsequent time period. We used Cox proportional hazards regression models to estimate hazard ratios (HRs) and 95% confidence intervals (95% CIs) in age- and multivariable-adjusted analyses. The HR is a measure of how often a particular event (e.g., new onset of tinnitus) happens in one group (e.g., frequent analgesic users) compared to how often it happens in another group (infrequent or non-users) over time. We used the Anderson-Gill data structure, 35 with a new data record created for each biennial questionnaire, to handle time-varying covariates efficiently. To control as finely as possible for confounding by age, calendar time, and any possible 2-way interactions between these two time scales, we stratified the analysis jointly by age in months at start of follow-up and calendar year of the current questionnaire cycle. Covariate status was determined at baseline and time-varying covariates were updated with each questionnaire cycle. Age, body mass index, physical activity, smoking, health conditions (e.g., hypertension, diabetes, depression, rheumatoid arthritis, systemic lupus erythematosus, other arthritis, migraine headache, hearing status) were updated every 2 years, caffeine and dietary intake every 4 years, and waist circumference and CCI score were updated when new covariate information was collected. The proportions of participants with missing data for each of the covariates at baseline, and overall, were relatively small. For example, the proportions of participants with missing information at baseline were ≤1% for aspirin, non-aspirin NSAIDs and acetaminophen, <0.1% for race, 4% for BMI, 6% for physical activity, <0.1% for smoking, and 12% for diet. For each of the analgesics, if a participant did not have information in a questionnaire cycle, the participant was skipped for that time period but was included in subsequent questionnaire cycles when the analgesic information was available. Missingness of the potential confounders was handled by the missing-indicator method. In view of our previous findings for analgesic use and hearing loss, we examined potential effect modification by age for each of the analgesics by including the interaction between age at the start of the current questionnaire cycle and the exposure.17 All P values are 2-tailed and considered statistically significant at P < 0.05. Statistical tests were performed with SAS statistical software, version 9.4 (SAS Institute Inc., Cary, NC).

Role of the Funding Source

This research was supported by grants U01 DC010811, R01 DC017717, and UO1 CA176726 from the National Institutes of Health. The funders had no role in the design, analysis, or conduct of the study or in the decision to submit the manuscript for publication.

Results

The baseline characteristics of the participants according to low (<1 day/week), medium (2–3 days/week), and high (6–7 days/week) frequency of use of aspirin, NSAID, and acetaminophen are shown in Table 1. Women who used any of the individual analgesics frequently were more likely to be current smokers and to have higher BMI, hypertension, diabetes, phobic anxiety, and depression. They were also more likely to report rheumatoid or other arthritis, and migraine headaches. Women who used either NSAIDs or acetaminophen frequently were more likely to use other analgesics regularly as well.

Table 1 Baseline (1995) Characteristics of Nurses’ Health Study II Participants, According to Analgesic Use. Analgesic Use Was Categorized According to Days/Week of Use: < 1, 2–3, 4–5, or 6–7 Days/Week. Three Categories for Illustration Purposes Are Displayed
Full size table

After 1,120,936 person-years of follow-up, 10,452 cases of incident persistent tinnitus were reported. Overall, more frequent analgesic use was independently associated with the risk of incident persistent tinnitus. The direction and the magnitude of the risk differed for the individual analgesics and are described individually below.

Aspirin

Overall, the risk of developing persistent tinnitus was not elevated among frequent users of low-dose or moderate-dose aspirin (Table 2). However, we found the associations for aspirin varied significantly by age (Table 3). For moderate-dose aspirin, frequent use was associated with higher risk of tinnitus among women aged < 60 years, but not among older women (p-interactionage = 0.003). Compared with women aged < 60 using moderate-dose aspirin < 1 day/week, the multivariable-adjusted hazard ratio (MVHR, 95% CI) among women using moderate-dose aspirin 6–7 days per week was 1.16 (1.03, 1.32).

Table 2 Analgesic Use and Risk of Incident Persistent Tinnitus Among Women in the Nurses’ Health Study II (1995–2017)
Full size table
Table 3 Low- and Moderate-Dose Aspirin Use and Risk of Incident Persistent Tinnitus Among Women in the Nurses’ Health Study II (2001–2017), Stratified by Age
Full size table

NSAIDs and COX-2 Inhibitors

More frequent use of NSAIDs was associated with higher risk of persistent tinnitus and the magnitude of the risk tended to be greater with increasing frequency of use (Table 2). Compared with women who used NSAIDs <1 day/week, the MVHRs were 1.07 (1.01, 1.13) among women who used NSAIDs 2–3 days/week, 1.17 (1.08, 1.28) for use 4–5 days/week, and 1.07 (1.00, 1.16) for use 6–7days/week (p-trend=0.001). The association between frequent NSAID use did not vary significantly by age (Table 4). For COX-2 inhibitors, regular use (2 or more days/week) was associated with higher risk of persistent tinnitus (Table 5). Compared with women who did not use COX-2 inhibitors regularly, the MVHR for use 2 or more days/week was 1.21 (1.10, 1.33).

Table 4 NSAID and Acetaminophen Use and Risk of Incident Persistent Tinnitus Among Women in the Nurses’ Health Study II (1995–2017), Stratified by Age
Full size table
Table 5 Regular Use of Selective COX-2 Inhibitors and Risk of Incident Persistent Tinnitus Among Women in the Nurses’ Health Study II (2001–2017). Regular use defined as 2 or more days per week
Full size table

Acetaminophen

More frequent use of acetaminophen was also associated with higher risk of persistent tinnitus and the magnitude of the risk tended to be greater with increasing frequency of use (Table 2). Compared with women who used acetaminophen <1 day/week, the MVHRs for persistent tinnitus were 1.00 (0.93, 1.08) among women who used acetaminophen 2–3days/week; 1.08 (0.96, 1.21) for use 4–5days/week; and 1.18 (1.07, 1.29) for use 6-7days/week (p-trend =0.002). Although the acetaminophen-age interaction was statistically significant (p= 0.02), the magnitudes of the association were only meaningfully different for use 4–5 days per week (Table 4).

In additional analyses, we categorized analgesic use according to the number of tablets per week for each of the analgesics and the multivariable-adjusted results were similar. In additional analyses that evaluated a more stringent definition of tinnitus, persistent tinnitus that lasted 5 min or longer and occurred every day, the results were not materially different for any of the analgesics (Supplemental Table 1). For example, compared with women who used NSAIDs <1 day/week, the MVHRs were 1.07 (1.00, 1.14) among women who used NSAIDs 2–3days/week; 1.17 (1.07, 1.29) for use 4–5days/week; and 1.10 (1.01, 1.20) for use 6–7days/week (p-trend =0.0007). For acetaminophen, compared with women who used acetaminophen <1 day/week, the MVHRs were 1.03 (0.94, 1.12) for use 2–3days/week; 1.13 (1.00, 1.34) for use 4–5days/week; and 1.20 (1.08, 1.34) for use 6–7days/week (p-trend =0.0006). Notably, 79% of participants who reported tinnitus that occurred several days per week or more also reported that their tinnitus lasted 5 min or longer.

Discussion

In this large longitudinal study among almost 70,000 women with over 20 years of follow-up, frequent analgesic use was independently associated with the risk of incident persistent tinnitus; the direction and the magnitude of the risk differed for the individual analgesics. Frequent use of low-dose (≤100 mg) aspirin was associated with a modest lower risk of tinnitus among women aged ≥60 years. In contrast, frequent use of moderate dose (≥325 mg) aspirin was associated with a 16% higher risk among younger women. Among all women, frequent use of NSAIDs or acetaminophen was associated with an almost 20% higher risk, and the magnitude of the elevated risk tended to be greater with more frequent use. Regular use of COX-2 inhibitors was also associated with a 21% higher risk. To our knowledge, this is first large longitudinal study of regular use of typical doses of these analgesics and risk of incident persistent tinnitus.

Tinnitus can be disabling and is one of the most common health conditions in the USA.36 A wide array of medications have been suggested to trigger or worsen tinnitus, generally at higher doses, and the tinnitus typically resolves after the medication is discontinued.37Non-prescription analgesics, such as aspirin, ibuprofen, and acetaminophen, are widely used in the USA8 and globally.7COX-2 inhibitors are available only by prescription in the USA and are prescribed to achieve the same analgesic and anti-inflammatory properties as NSAIDs, but with less GI toxicity.10, 11

The effects of aspirin and the respective mechanisms of action vary with dose. High-dose aspirin is associated with transient reductions in otoacoustic emissions, cochlear threshold shifts, and reduced cochlear neural output.15 Functional imaging studies suggest aspirin may increase spontaneous activity in central auditory pathways and functional connectivity in auditory and non-auditory brain regions, including those associated with arousal and emotion, such as the reticular formation and amygdala,14, 15 and possibly mediated by alterations of γ-aminobutyric acid (GABA) and serotonin-mediated neurotransmission.38

The dose-dependent association between aspirin and incident tinnitus has been demonstrated in animals and in humans.15 We did not observe an elevated risk among frequent low-dose aspirin users. The significant, yet modest, protective benefit of low-dose aspirin use among older women was unexpected. Low dose (≤100 mg) aspirin produces a clinically relevant antiplatelet effect by irreversibly acetylating the active site of cyclooxygenase-1 (COX-1), which is required for the production of thromboxane A2, a powerful promoter of aggregation. Low-dose aspirin use is associated with reduced risk of cardiovascular events39 and stroke.40 The potential beneficial vascular effects of low-dose aspirin may be influential in preserving adequate cochlear blood flow with advancing age, as the consequences of factors such as atherosclerosis and endothelial dysfunction accumulate and increase the risk of compromised cochlear blood flow and resultant hypoxia, cell injury, and reduced cochlear output. Nevertheless, the inverse association among older women was modest; more importantly, the risk was not elevated with frequent low-dose aspirin use.

Conversely, moderate-dose aspirin was associated with higher risk, particularly among younger women. Moderate doses of aspirin also inhibit COX-2, which blocks prostaglandin production leading to analgesic and antipyretic effects. There are also several prostaglandin-independent actions of aspirin, including inhibition of neutrophil activation and responses, inhibition of inducible nitric oxide synthase expression and partial suppression of nitric oxide generation, and inhibition of activation of certain factors involved in inflammation and infection, including interleukin-1, interleukin-6, tumor necrosis factor, adhesion molecules, and nuclear factor-kappa B.41,42,43 Our finding that frequent use of moderate doses of aspirin when taken for longer duration was associated with elevated risk of persistent tinnitus could be mediated by mechanisms similar to those for high-dose aspirin. Advancing age is a strong risk factor for cochlear dysfunction and hearing loss. Possibly, the influence of moderate-dose aspirin on cochlear function among younger individuals with less pre-existing cochlear dysfunction may be greater with respect to its ability to induce tinnitus. That is, in the setting of pre-existing cochlear dysfunction, the additional impairment due to moderate-dose aspirin use may not lead to alterations in cochlear output of sufficient magnitude to increase central gain and result in tinnitus.

In addition, age-related differences in the response to moderate-dose aspirin may reflect aging-related differences in neural plasticity. Evidence suggests that long-term maintenance of tinnitus is likely a function of neuroplastic changes in a complex network of structures involving central auditory and nonauditory pathways.44, 45 Salicylates cross the blood-brain barrier and have been shown to increase sound-evoked neural activity along central auditory pathways, possibly due to disruptions of inhibitory circuits.15Age-related differences in auditory neural plasticity have been demonstrated.46, 47 Further, in rats, aspirin administration in doses sufficient to induce tinnitus behaviors also resulted in substantially increased plasma corticosterone levels in a dose-dependent manner.48 Potentially, age-related differences in the relative increases in cortisol associated with moderate-dose aspirin use49 may in turn differentially predispose younger versus older individuals to the development of tinnitus. The age- and dose-related differences in the association of aspirin use and risk of developing tinnitus are intriguing and merit further study.

NSAIDs exert analgesic, antipyretic, and anti-inflammatory actions, predominantly through the inhibition of cyclo-oxygenase(COX) and reduced transformation of arachidonic acid to prostaglandins, prostacyclin, and thromboxanes.50, 51 Two related isoforms of the COX enzyme, COX-1 and COX-2, differ in their regulation and expression in various tissues.10 The degree of enzyme inhibition by different NSAIDs varies, and the extent to which a particular NSAID inhibits a COX isoform influences its activity and toxicity.11 NSAIDs may impair prostaglandin-mediated cochlear vasodilation and lead to compromised cochlear blood flow and function.12Non-prostaglandin-mediated mechanisms of NSAIDs include disruption of cell membrane function and inhibition of neutrophil function, which may alter the response to inflammation.52 Notably, use of NSAIDs in typical doses 5 or more days per month was associated with higher risk of hypertension,22 and use of NSAIDs at high frequency or doses was associated with increased risk for major cardiovascular events.40 Use of very high doses of NSAIDs has been associated with inner ear perturbations and transient reversible tinnitus and hearing loss.12 Adverse auditory effects of high-dose non-salicylate NSAIDs may be similar to those of high-dose salicylates; however, the influence of typical dose NSAIDs on auditory function is not known. Frequent use and longer duration of use of NSAIDs were previously shown to be associated with higher risk of hearing loss.17,18,19 Although a cross-sectional association between prevalent tinnitus and current NSAID use was found,53 to our knowledge, this is the first longitudinal study to evaluate typical dose non-salicylate NSAID use and risk of incident tinnitus.

Acetaminophen is one of the most commonly used medications. Acetaminophen and acetaminophen-containing medications represent a majority share in the over-the-counter analgesic market,54 and in a given week it is used by approximately 20% of the US adult population.8 Moreover, acetaminophen is an active ingredient in hundreds of over-the-counter and prescription medications indicated for pain, fever, colds, flu, allergies, and sleeplessness.9

Acetaminophen use is associated with an increased risk of hypertension,55,56,57,58 asthma59, and cardiovascular events.40 In the ear, acetaminophen may deplete glutathione and render the cochlea vulnerable to oxidative damage,13, 60 which may increase the risk of tinnitus. Evidence suggests reduced glutathione S-transferase detoxification may increase susceptibility to age- and noise-related hearing loss.13 Although only weakly inhibiting COX-1 and COX-2,61 acetaminophen also inhibits prostaglandin production62 and may impair endothelial function through depletion of glutathione.63 Acetaminophen may also mediate some of its effects through inhibition of a splice variant of COX-1, COX-3, which is present in the cerebral cortex.61 We previously found that frequent use of acetaminophen was associated with higher risk of hearing loss.17, 18

Our study has several strengths, including its large size and long duration of follow-up. Most previous studies of factors associated with tinnitus have been cross-sectional; few studies have examined risk factors for incident tinnitus and to our knowledge none have assessed tinnitus incidence over more than 10 years. We collected detailed information on analgesics from a large number of participants, permitting investigation of a broad range of intake. We asked distinct questions about low- and moderate-dose aspirin, NSAIDs, acetaminophen, and COX-2 inhibitors; thus, we were able to examine these drugs individually. We obtained analgesic data prospectively, before onset of tinnitus. To account for changes in participants’ patterns of use over several years, we updated analgesic data biennially and used the most recent exposure and risk factor data for each 2-year time period. Notably, our cohort of nurses likely provides reliable data on consumption of prescription and over-the-counter medications. Several limitations deserve comment. Information on tinnitus and analgesic use was self-reported. Due to the nature of the condition, subjective tinnitus is perceived only by the individual; therefore, the diagnosis of tinnitus must rely on self-report.64 Although we do not have information on whether participants sought medical attention for their tinnitus, it is notable that in a study using data from the National Health Interview Survey (N=75,764), Bhatt et al. found that among those individuals who experienced tinnitus, only 50.6% had discussed their tinnitus with a health care provider.36 Notably, among those who reported persistent tinnitus in 2009, 88% also reported persistent tinnitus again 4 years later on the 2013 questionnaire. Definitions of tinnitus used in previous epidemiologic studies have varied greatly;27 however, when we examined alternative definitions of tinnitus in additional analyses, the results were similar. Among individuals with tinnitus, there is considerable heterogeneity in the etiology, pathophysiology, and clinical characteristics. The observational design of our study does not permit us to assign causality. Although residual confounding could have influenced the results, the analyses were carefully adjusted for potentially confounding variables, many of which were demonstrated to be well-reported in this cohort. We did not have drug prescription information on our participants; however, the vast majority of the analgesics used in our cohort are available over-the-counter. Medication use was self-reported, but the validity and reliability of the self-reported information collected in this and similar cohorts on a wide array of health-related, dietary, and lifestyle factors using these methods have been demonstrated in a number of previously published validation studies using information from daily/weekly diaries, medical records, and biologic samples.65,66,67 In addition, this assessment of analgesic use in NHS II and related cohorts has been used in numerous studies and important associations have been identified.17, 18, 68, 69 We did not have information on reasons why women used the specific analgesics, but we adjusted for numerous factors that have been reported to be the most common indications for analgesic use. Nevertheless, the possibility of residual confounding remains. We did not have information on the specific NSAID formulation; thus, we were unable to study individual NSAIDs. However, in a previous study, we found that ibuprofen was the most commonly used NSAID, being used by ~80% of women taking an NSAID. We did not have information on specific dose of NSAIDs or acetaminophen, but in additional analyses that examined analgesic intake according to the number of tablets taken per week, the results did not differ. The study population included predominantly White female health care professionals; thus, research in non-White women and additional populations is warranted.

Conclusion

In this large longitudinal study of analgesic use and risk of incident persistent tinnitus, the risk of developing persistent tinnitus was not elevated among frequent low-dose aspirin users. Among younger women, frequent moderate-dose aspirin use was associated with higher risk. Frequent NSAID use and frequent acetaminophen use were associated with higher risk of incident persistent tinnitus among all women, and the magnitude of the risks tended to be greater with increasing frequency of use. Our results suggest analgesic users are at higher risk for developing tinnitus and may provide insight into the precipitants of this challenging disorder, but additional investigation to determine whether there is a causal association is needed. Further, these findings underscore the importance of careful evaluation of ongoing use of over-the-counter and prescription analgesic medications, weighing their benefits with potential risks.