You have accessEditor's AwardJournal of Speech, Language, and Hearing ResearchResearch Article8 Nov 2021

Blast Exposure and Self-Reported Hearing Difficulty in Service Members and Veterans Who Have Normal Pure-Tone Hearing Sensitivity: The Mediating Role of Posttraumatic Stress Disorder

    Abstract

    Purpose

    Evidence suggests that military blast exposure may lead to self-reported hearing difficulties despite audiometrically normal hearing. Research identifying potential mechanisms of this association remains limited. The purpose of this article is to evaluate the associations between blast, posttraumatic stress disorder (PTSD), and self-reported hearing difficulty, and to examine PTSD as a possible mediator of the association between blast exposure and hearing difficulty.

    Method

    We used baseline data from the Noise Outcomes in Service members Epidemiology (NOISE) study (n = 477). Participants in this study undergo a comprehensive hearing, and tinnitus if applicable, evaluation and complete a large number of surveys. Pertinent data extracted from these surveys included information on participant's demographics, military service history, including exposure to blast, and health conditions such as symptoms of PTSD. Using regression models and following a formal causal mediation framework, we estimated total associations, natural direct and indirect associations, and percent mediated.

    Results

    We found that individuals with blast exposure had higher prevalence of both probable PTSD and self-reported hearing difficulty than individuals who were not blast exposed. Compared with participants without blast exposure, those with blast exposure had twice the prevalence of self-reported hearing difficulty, with 41% of the association mediated through probable PTSD.

    Conclusion

    As PTSD is a possible mediator of the association between blast exposure and hearing difficulty, Service members and Veterans with normal pure-tone hearing sensitivity who report hearing difficulties and a history of blast exposure may benefit from evaluation for PTSD symptoms.

    Supplemental Material

    https://doi.org/10.23641/asha.16674247

    Blast exposure is a known risk factor for hearing loss, especially in the military population (Elsayed, 1997; Institute of Medicine [IOM], 2014; Joseph et al., 2018). Increased insurgence warfare in post–9/11 conflicts has led to an increase in blast exposures among military Service members compared with past wars (IOM, 2014). Thanks to advanced personal protective equipment and combat casualty care capabilities, today's Service members are more likely to survive blast-related injuries. This progress means that more Service members return home alive, but many suffer with lasting deficits and disabilities (IOM, 2014; Rafaels et al., 2011). The ear is particularly vulnerable to a blast pressure wave because it is the body's most sensitive pressure transducer (Elsayed, 1997; Mathews & Koyfman, 2015). In a sample of deployed male Service members, Joseph et al. (2018) found that individuals with a blast-related injury were twice as likely to have peripheral hearing loss compared with individuals with a non–blast-related injury (adjusted odds ratio = 2.2; 95% CI [1.4, 3.4]). Forty-nine percent of the excess peripheral hearing loss was attributed to the blast event. Although individuals with peripheral hearing loss are likely to report increased difficulty understanding speech in noise and in other complex acoustic environments, individuals with normal or near-normal hearing sensitivity have also reported hearing difficulty (Bressler et al., 2017; Gallun, Diedesch, et al., 2012; Gallun et al., 2016; Saunders et al., 2015).

    Veterans who demonstrate normal or near-normal pure-tone hearing sensitivity have reported hearing difficulties when trying to understand the content of speech in the presence of noise, when attempting to follow long conversations, and when listening by telephone (Saunders et al., 2015). Among blast-exposed Veterans with normal to near-normal pure-tone hearing sensitivity, 60% self-reported difficulty hearing compared to only 7% without blast exposure (Gallun et al., 2016). One theorized mechanism of blast-related hearing injury describes the blast wave propagating through the brain, resulting in contusions, shearing, and diffuse axonal injury within central auditory structures (Gallun, Lewis, et al., 2012; Gallun et al., 2017). Blast exposure is the most common cause of traumatic brain injury (TBI) among U.S. forces deployed to Iraq and Afghanistan (IOM, 2014). Another theorized mechanism of blast-related hearing injury is through subclinical cochlear hearing loss undetected by traditional hearing tests (Bressler et al., 2017; Oleksiak et al., 2012). The ability to understand speech in noise and in other complex acoustic environments also requires sustained attention and focus on the target speech signal. This involves cognitive functions that remain mostly outside auditory structures.

    Two pathologies that are not specific to the auditory system, are relevant to a discussion of blast exposure, and may interfere with sustained attention and focus are posttraumatic stress disorder (PTSD) and TBI (Vasterling et al., 1998; Vasterling et al., 2002). PTSD is a mental health condition that is triggered by a traumatic event along with a reaction of fear, helplessness, or horror to the event (American Psychiatric Association [APA], 2013). Symptoms of PTSD may include re-experiencing the event, avoidance, and changes in physical and emotional reactions such as hyperarousal (APA, 2013). Individuals with PTSD often describe difficulties with concentration, attention, and memory (Dolan et al., 2012; Vasterling et al., 1998). Veterans with PTSD have shown deficits on tasks that require working memory, sustained attention, and processing speed (Dolan et al., 2012; Vasterling et al., 2002). Considering the well-established relationship between military blast exposure and PTSD (IOM, 2014), it is possible that effects of blast on hearing difficulty may also be mediated by PTSD. No one has previously studied the role of PTSD as a possible mediator between blast exposure and hearing sequelae. Relevant studies have only expressed the high co-occurrence of blast exposure and PTSD, noting the marked complexity in understanding the contributions of PTSD to the observed association (Bressler et al., 2017; Gallun, Diedesch, et al., 2012; Gallun et al., 2016; Saunders et al., 2015; Tepe et al., 2020). This points to the need for a unifying conceptual framework for blast-related functional hearing difficulties that integrates PTSD.

    TBI is often categorized as mild (also referred to as “concussion”), moderate, or severe. Approximately 80% of all military TBIs are categorized as mild (IOM, 2014; Terrio et al., 2009). Evidence connecting mild TBI to TBI-related symptoms and sequelae is inconsistent and is often attributable to PTSD symptoms (Carlson et al., 2009; Vasterling et al., 2018). It is not surprising then that these two pathologies frequently co-occur; 64% of Iraq and Afghanistan war Veterans with a diagnosis of mild TBI also have a co-morbid PTSD diagnosis (Carlson et al., 2010). As mentioned previously, the most common cause of TBI among U.S. forces deployed to Iraq and Afghanistan is blast exposure, which can also result in PTSD (IOM, 2014). For these reasons, and because PTSD symptoms are potentially modifiable, it is important to recognize and assess PTSD as a possible mediator of sensory sequalae linked to blast exposure.

    The primary aim of this study was to examine, in a group of Service members and Veterans, the associations between blast, PTSD, and self-reported hearing difficulty, and secondarily, to examine PTSD as a possible mediator of the association between blast exposure and hearing difficulty. An improved understanding of mechanisms that mediate an association imparts support of the primary association while also highlighting a new potential target for intervention. We build upon past examinations by proposing an a priori causal relationship between blast, PTSD, and self-reported hearing difficulty as depicted in our conceptual model (see Figure 1). The conceptual model highlights five epidemiologic questions that motivated our study: (a) What is the effect of blast on PTSD? (b) What is the effect of PTSD on self-reported hearing difficulty? (c) What is the total effect of blast on self-reported hearing difficulty? (d) What is the direct effect of blast on self-reported hearing difficulty (i.e., the effect not mediated through PTSD)? and (e) What is the indirect effect of blast on self-reported hearing difficulty mediated through PTSD?

    Figure 1.

    Figure 1. Conceptual model identifying the theorized causal relationships between blast and functional hearing difficulty. The dashed line implies effect measure modification (i.e., posttraumatic stress disorder [PTSD] modifies the effect measure between blast exposure and hearing difficulty). Potential confounders are grouped by demographics and service characteristics. A more detailed conceptual model can be found in Supplemental Material S1.

    Method

    Participants were members of The Noise Outcomes in Service members Epidemiology (NOISE) study, an ongoing longitudinal study designed to examine the effects of military and nonmilitary exposures on auditory functioning among post-9/11 Service members and Veterans (Henry et al., 2021). Study recruitment and follow-up are ongoing. The NOISE study cohort includes active duty Service members and Veterans who have recently (within approximately 2.5 years) separated from the military. Individuals are enrolled into the NOISE study at the VA Rehabilitation Research & Development Service National Center for Rehabilitative Auditory Research at the VA Portland Health Care System and at the Department of Defense's Hearing Center of Excellence at Joint Base San Antonio, Texas. Potential study participants are excluded if they did not serve post-9/11 or if they have been separated for more than 2.5 years from military service. Study participants undergo a comprehensive hearing evaluation (otoscopy, air- and bone-conduction audiometry, speech audiometry, acoustic immittance, and otoacoustic emissions) and, if tinnitus is present, a tinnitus psychoacoustic evaluation. Study participants also complete a large battery of surveys. Surveys query individuals with respect to demographics; military characteristics; exposures to noise, ototoxic agents, and ototraumatic events; perceived hearing health, mental health, and physical health; and quality of life. Further details on the NOISE study design and cohort can be found elsewhere (Gordon et al., 2017; Henry et al., 2021). Informed consent was obtained from participants prior to data collection, and they were compensated for their effort. This study received approval and oversight from the respective Institutional Review Boards at each study location—the VA Portland Health Care System (#3159)/Oregon Health and Science University (#9495) Joint Institutional Review Board and the Medical Research and Materiel Command Institutional Review Board (M-10423).

    Data collected during the baseline evaluation among individuals enrolled in the NOISE study between 2014 and 2019 were used in this analysis. The sample for this analysis was limited to NOISE study participants who had normal pure-tone hearing sensitivity bilaterally, which we defined as hearing thresholds ≤ 20 dB hearing level (dB HL) from 250 to 8000 Hz. We excluded individuals with a history of moderate or severe TBI from our analysis. Moderate to severe TBI likely yields a different conceptual model than presented in Figure 1. Relevant data extracted from NOISE study surveys used in the analysis are described below.

    Exposure Measurement: Blast

    The Blast and Traumatic Brain Injury Questionnaire was designed specifically for our study to assess participant history of probable TBI and blast exposures. This questionnaire was adapted from the Comprehensive TBI Evaluation used clinically by the VA (Veterans Health Administration, 2017) to include questions related to TBI- and blast-induced acoustic injury and tinnitus. To capture information about blast exposure, participants were asked: “Now we are going to ask you about your blast exposures, whether or not you experienced any injuries related to blasts. When a high explosive bomb or improvised explosive device (IED) goes off, there is a ‘blast wave,’ which is a wave of highly compressed gas that hits solid objects like a person's body and may feel almost like smashing into a wall. Do you remember experiencing this type of ‘blast wave’ or ever being told that you experienced it?” Response options are “yes,” “not sure,” or “no.” For the current analysis, blast exposure was coded dichotomously (yes/no) with all responses of “unsure” recoded as “no.”

    Mediator Measurement: PTSD

    PTSD was measured using the Primary Care-PTSD-4 screening questionnaire (Prins et al., 2003). The PTSD-4 is a four-item questionnaire used in primary care and other clinical settings at the VA. The screening questionnaire begins with a sentence to cue study participants to consider traumatic events. Participants noting a traumatic event are asked if that event was so upsetting that over the last month they (a) had nightmares about the event, (b) avoided situations that reminded them of the event, (c) were “on guard,” or (d) felt detached from others and their surroundings. If responses were “yes” to at least three items, they were considered to have probable PTSD. The PTSD-4 has been shown to have high sensitivity (0.78) and specificity (0.87) when compared with a PTSD diagnosis in a Veteran primary care population (Prins et al., 2003).

    Outcome Measurement: Self-Reported Hearing Difficulty

    Self-reported hearing difficulty was captured using the Hearing Handicap Inventory for Adults (HHIA). The HHIA is a 25-item hearing questionnaire designed to measure the emotional and social adjustment of people with perceived hearing difficulty (Newman et al., 1990). It can be used regardless of whether peripheral hearing loss is present. Responses to the items include “yes” (4 points), “sometimes” (2 points), and “no” (0 points). Scores range from 0 to 100, with higher scores indicating greater hearing difficulties. The HHIA was coded dichotomously without differentiation between social and emotional impacts; based on the original Hearing Handicap Inventory for the Elderly, from which the HHIA was derived, total scores 0–16 suggest no difficulty and scores >16 suggest the presence of functional hearing difficulty (Ventry & Weinstein, 1982).

    Observed Covariates

    Potential confounders were selected a priori using directed acyclic graphs (Greenland et al., 1999; Hernán et al., 2002); these were informed by theory and prior knowledge (see Supplemental Material S1). All potential confounders were self-reported. We obtained them from our Demographic Questionnaire. These confounders included gender (Men; Women), age (in years), race (White; African American/Black; American Indian/Alaskan Native; Asian; Hawaiian/Pacific Islander; Other or Selected > 1 Race; Prefer Not To Answer/Missing), ethnicity (non-Spanish/Hispanic/Latino; Spanish/Hispanic/Latino), marital status (Living with Spouse/Partner; Single/Never Married; Divorced/Separated/Widowed), education (Some/Completed High School; Some College Vocational; Completed College), service branch (Army; Marine Corps; Navy/Coast Guard; Air Force), duration of service (in years), and deployment to a conflict zone (Yes; No). In the analysis of effects of PTSD on hearing difficulty, blast exposure was also included as a potential confounder. If a Service member or Veteran had enlisted in more than one service branch, then service branch was assigned based on longest duration of service.

    Statistical Analysis

    Our sample consisted of 490 Service members and Veterans who had pure-tone hearing sensitivity within normal limits. Thirteen participants (2.7%) had one or more missing variables and were excluded from the analysis (final n = 477). All analyses were conducted using SAS software, version 9.4.

    Unadjusted and Adjusted Total Effects Analysis

    Covariate distribution was examined among study participants with and without blast (exposure) as well as with and without probable PTSD (mediator). Bivariable and multivariable log-binomial models were used to first estimate prevalence proportion ratios (PPRs) and 95% confidence intervals (CIs) describing the association between blast exposure and self-reported hearing difficulty and between blast exposure and probable PTSD. Next, bivariable and multivariable log-binomial models were used to estimate PPR and 95% CI describing the association between probable PTSD and self-reported hearing difficulty, including adjustment for blast exposure (a confounder of the PTSD/self-reported hearing difficulty association). Unadjusted models and models adjusted for demographic and service characteristics are presented. Due to small cell sizes, race and ethnicity were collapsed into a single variable reflecting non–Hispanic White study participants versus all other races/ethnicities for inclusion in the adjusted models. We conceptualized this covariate as an indicator of discrimination and racism rather than as a biologic risk factor, given racism likely structures adverse military exposures and health outcomes (Boyd et al., 2020). We assessed the sensitivity of our adjusted regression models to potential unmeasured confounding by computing an E-value; the E-value is an estimation of strength of association between an unmeasured confounder and exposure, as well as outcome, necessary to suggest a reasonable alternative explanation for the observed association (VanderWeele & Ding, 2017).

    Mediation Analysis, Accounting for Exposure/Mediator Interaction

    The secondary aim was to partition the total effect between blast exposure and self-reported hearing difficulty into natural direct effects (NDEs) and natural indirect effects (NIEs; Hafeman & Schwartz, 2009; Petersen et al., 2006). The NDE is the effect of blast exposure on self-reported hearing difficulty if the effect of the exposure on the mediator (PTSD) had somehow been blocked. Alternatively, the NIE is the effect of blast exposure on self-reported hearing difficulty that would be realized if probable PTSD were somehow changed to what it would be without blast exposure. We used mediation analysis methods based on a regression framework for a common dichotomous outcome (Ananth & VanderWeele, 2011; Spiegelman & Hertzmark, 2005). The NDE and NIE of the blast–hearing difficulty relationship were estimated by fitting a log-binomial model for self-reported hearing difficulty (Y), conditional on blast exposure (A), probable PTSD (M), a Blast × PTSD interaction variable (AxM), and a set of confounders (C): log{P(Y = 1|A = a,M = m,C = c)} = θ0 + θ1a + θ2m + θ3am + θ′4c, as well as a logit model for probable PTSD, conditional on blast exposure and confounders: logit{P(M = 1|A = a,C = c)} = β0 + β1a + β2c.(Valeri & VanderWeele, 2013). The proportion of the increased blast exposure, self-reported hearing difficulty association that is mediated (i.e., proportion mediated) through probable PTSD was computed on the prevalence proportion difference scale as Proportion Mediated = PPR NDE PPR NIE 1 / PPR NDE PPR NIE 1 100 (Ananth & VanderWeele, 2011). Four-way decomposition was used to disentangle the association between blast exposure and hearing difficulty by probable PTSD, explaining “how” and “for whom” a cause affects an outcome (VanderWeele, 2014). Four-way decomposition includes estimating the following effects: the controlled direct effect (CDE), reference interaction, mediated interaction, and the pure indirect effect (PIE). In this analysis, the CDE estimates the effect of blast exposure on self-reported hearing difficulty that does not go through PTSD. The reference interaction estimates the effect of the interaction between blast exposure and PTSD on self-reported hearing difficulty. The mediated interaction effect is the interaction of blast exposure with PTSD that is likely caused by blast exposure. Finally, the PIE estimates the effect due to mediation by PTSD alone, without interactions. Standard errors were obtained using the delta method. E-values were estimated for our causal mediation analysis to examine the potential impact of an unmeasured confounder on our results. This is a cross-sectional analysis of baseline data; we use the term “association” rather than “effect” to describe the results when appropriate while retaining the causal mediation nomenclature for clarity (e.g., NDE, NIE, CDE, and PIE).

    Results

    Of the 477 Service members and Veterans with normal pure-tone hearing sensitivity, 113 (23.7%) reported blast exposure, 109 (22.9%) screened positive for probable PTSD, and 100 (21.0%) reported hearing difficulty, according to the HHIA. Forty-nine (10.3%) participants reported having blast exposure as well as screened positive for probable PTSD. The median level of self-reported hearing difficulty for all 477 subjects was 4 (HHIA range: 0–96). Distribution of covariates by blast history (exposure) and probable PTSD (mediator) for the study sample are shown in Table 1. The average age in the sample was about 31 years and varied little by blast exposure or by probable PTSD. Blast exposure and probable PTSD was experienced by more men than women participants and more by participants reporting combat deployment than those without combat deployment (see Table 1). There were no observable differences in mean hearing thresholds between those with and without blast exposure or with and without probable PTSD.

    Table 1. Distribution of covariates by blast (exposure) and probable posttraumatic stress disorder (PTSD; mediator) for 477 study participants. N (%) are displayed unless otherwise noted.

    Characteristic Blast exposure
    Probable PTSD
    Total
    Yes
    No
    Yes
    No
    n (%) n (%) n (%) n (%) n
    Gender
     Men 98 (29.8) 231 (70.2) 82 (24.9) 247 (75.1) 329
     Women 15 (10.1) 133 (89.9) 27 (18.2) 121 (81.8) 148
    Age years; mean (SD) 32 (6.9) 31 (7.3) 31 (6.9) 31 (7.3) 477
    Hearing 0.25–2 kHz average
     Right ear; mean (SD) 9.9 (4.2) 8.4 (3.8) 9.9 (3.9) 8.4 (3.9) 477
     Left ear; mean (SD) 9.8 (4.6) 8.3 (4.0) 10.1 (4.2) 8.2 (4.1) 477
    Hearing 3–8 kHz average
     Right ear; mean (SD) 8.3 (4.5) 6.4 (4.2) 7.8 (4.1) 6.6 (4.4) 477
     Left ear; mean (SD) 8.0 (4.6) 6.3 (4.3) 7.7 (4.1) 6.4 (4.5) 477
    Ethnicity
     Non-Spanish/Hispanic/Latino 91 (22.5) 314 (77.5) 95 (23.5) 310 (76.5) 405
     Spanish/Hispanic/Latino 19 (27.9) 49 (72.1) 11 (16.2) 57 (83.8) 68
     Prefer not to answer/missing 3 (75.0) 1 (25.0) 3 (75.0) 1 (25.0) 4
    Race/ethnicity
     Non-Hispanic White 70 (24.4) 217 (75.6) 67 (23.3) 220 (76.7) 287
     All other races/ethnicities 43 (22.6) 147 (77.4) 42 (22.1) 148 (77.9) 190
    Marital status
     Living with spouse/partner 70 (26.1) 198 (73.9) 63 (23.5) 205 (76.5) 268
     Single, never married 24 (16.1) 125 (83.9) 32 (21.5) 117 (78.5) 149
     Divorced/separated/widowed 19 (31.7) 41 (68.3) 14 (23.3) 46 (76.7) 60
    Education
     Some/completed high school 10 (21.7) 36 (78.3) 13 (28.3) 33 (71.7) 46
     Some college/vocational 71 (26.4) 198 (73.6) 72 (26.8) 197 (73.2) 269
     Completed college 32 (19.8) 130 (80.2) 24 (14.8) 138 (85.2) 162
    Military branch
     Army 62 (40.5) 91 (59.5) 48 (31.4) 105 (68.6) 153
     Marine Corps 19 (33.9) 37 (66.1) 21 (37.5) 35 (62.5) 56
     Navy/Coast Guard 9 (13.4) 58 (86.6) 17 (25.4) 50 (74.6) 67
     Air Force 23 (11.4) 178 (88.6) 23 (11.4) 178 (88.6) 201
    Military component
     Active 90 (21.4) 331 (78.6) 92 (21.9) 329 (78.1) 421
     National Guard/Reserve 23 (41.1) 33 (58.9) 17 (30.4) 39 (69.6) 56
    Duration years; median (range) 8.0 (2–28) 6.6 (0.2–31) 7.2 (1.5–28) 7.0 (0.2–31) 477
    Combat deployment
     Yes 106 (35.8) 190 (64.2) 88 (29.7) 208 (70.3) 296
     No 7 (3.9) 174 (96.1) 21 (11.6) 160 (88.4) 181
    Probable history of mild TBI
     Yes 39 (68.4) 18 (31.6) 22 (38.6) 35 (61.4) 57
     No 74 (17.6) 346 (82.4) 87 (20.7) 333 (79.3) 420

    Note. Row percentages are shown. SD = standard deviation; kHz = kilohertz; TBI = traumatic brain injury.

    Total Effects Analysis

    Prevalence proportion (%), unadjusted and adjusted PPRs, and the 95% CI of the observed associations are presented in Table 2. Participants with blast exposure had a higher prevalence of self-reported hearing difficulty and a higher prevalence of probable PTSD compared with those without blast exposure. After adjusting for potential confounders, those with blast exposure had 2.0-fold (95% CI [1.3, 3.0]) higher prevalence of self-reported hearing difficulty (see Table 2A) and 1.8-fold (95% [1.3, 2.6]) higher prevalence of probable PTSD (see Table 2B), compared to those without blast exposure.

    Table 2. Proportion of study participants with hearing difficulty and probable posttraumatic stress disorder (PTSD) by blast exposure (A and B, respectively) and hearing difficulty by probable PTSD (C).

    A. Exposure–Outcome
    Blast Hearing difficulty
    Unadjusted
    Adjusteda
    Yes No PPR 95% CI PPR 95% CI
    Yes 43 (38.1%) 70 (61.9%) 2.4 [1.7, 3.4] 2.0 [1.3, 3.0]
    No 57 (15.7%) 307 (84.3%) 1.0 Ref 1.0 Ref

    B. Exposure–Mediator
    Probable PTSD
    Unadjusted
    Adjusted
    a
    Blast
    Yes
    No
    PPR
    95% CI
    PPR
    95% CI
    Yes 49 (43.4%) 64 (56.6%) 2.6 [1.9, 3.6] 1.8 [1.3, 2.6]
    No 60 (16.5%) 304 (83.5%) 1.0 Ref 1.0 Ref

    C. Mediator–Outcome
    Hearing difficulty
    Unadjusted
    Adjusted
    b
    Probable PTSD
    Yes
    No
    PPR
    95% CI
    PPR
    95% CI
    Yes 53 (48.6%) 56 (51.4%) 3.8 [2.7, 5.3] 3.1 [2.2, 4.3]
    No 47 (12.8%) 321 (87.2%) 1.0 Ref 1.0 Ref

    Note. Unadjusted and adjusted total associations between blast–hearing difficulty (A, Exposure–Outcome), blast–probable PTSD (B, Exposure–Mediator), and probable PTSD–hearing difficulty (C, Mediator–Outcome,) as estimated from log-binomial regression. PPR = prevalence proportion ratio; CI = confidence interval.

    aAdjusted for gender, age, race/ethnicity, education, marital status, service branch, service component, service duration, and combat deployment.

    bAdjusted for the above (a) plus blast exposure.

    Additionally, 49% of participants with probable PTSD versus 13% of those without probable PTSD reported difficulty hearing. After adjusting for potential confounders, including blast exposure, those with probable PTSD had 3.1-fold (95% CI [2.2, 4.3]) higher prevalence of self-reported hearing difficulty than those without probable PTSD (see Table 2C). Further adjustment for mild TBI did not appreciably change this finding (data not shown). E-values suggest that moderate confounding would be required to explain away the observed associations (i.e., exposure/confounder and outcome/confounder relative risk > 3.0; see Table 4).

    Mediation Analysis Results

    The total blast–self-reported hearing difficulty association was partitioned into NDE and NIE mediated through PTSD (see Table 3). The total association was attenuated after adjustment for demographic and service characteristics; despite adjustment, the point estimate remained elevated. The blast–hearing difficulty association was slightly larger for NDE than NIE. Results for the NDE (the effect that would be realized if the effect of blast exposure on PTSD had been blocked) showed an association with hearing difficulty (NDE = 1.7; 95% CI [1.1, 2.3]). Results for the NIE (the effect that would be realized if the mediator, probable PTSD, were somehow changed to what it would be without blast exposure) also showed an association with hearing difficulty (1.3; 95% CI [1.0, 1.5]). We estimate that just over 41% (95% CI [17, 66%]) of the total association was mediated through probable PTSD. The estimated E-values were 2.8 and 1.9, suggesting that moderate confounding would be required to explain away the observed NDE and NIE, respectively (see Table 4).

    Table 3. Causal mediation model results demonstrating the natural direct effect of blast exposure on hearing difficulties unmediated by posttraumatic stress disorder (PTSD), the natural indirect effect mediated by PTSD, and proportion of the relationship between blast exposure and self-reported hearing difficulty mediated by PTSD.

    Effect Unadjusted
    Adjusteda
    PPR 95% CI PPR 95% CI
    Total effect 2.4 [1.6, 3.2] 2.2 [1.4, 3.0]
    Natural direct effect 1.8 [1.1, 2.5] 1.7 [1.1, 2.3]
    Natural indirect effect 1.4 [1.1, 1.6] 1.3 [1.0, 1.5]

    Proportion mediatedb

    45%

    [18, 72%]

    41%

    [17, 66%]

    Note. Unadjusted and adjusted models and their 95% confidence interval (CI) are displayed. PPR = prevalence proportion ratio.

    aAdjusted for gender, age, race/ethnicity, education, marital status, service branch, service component, service duration, and combat deployment.

    bProportion mediated = NDE NIE 1 / NDE NIE 1 100.

    Table 4. E-values for the point estimate and lower confidence interval for the fully adjusted regression models and the causal mediation model.

    Model E-value for point estimate E-value for lower confidence interval
    Fully adjusted regression models
     Blast ➔ Hearing difficulty 3.41 1.92
     Blast ➔ PTSD 3.00 1.92
     PTSD ➔ Hearing difficulty 5.65 3.82
    Causal mediation model
     Total effect 3.82 2.15
     Natural direct effect 2.79 1.43
     Natural indirect effect 1.92 1.00

    Note.  The E-value represents the strength of an unmeasured confounder necessary to suggest a reasonable alternative explanation for the observed associations (VanderWeele & Ding, 2017). PTSD = posttraumatic stress disorder.

    Results of the four-way decomposition analysis are reported in Table 5, which provides further insight into the composition of the total association between blast exposure and self-reported hearing difficulty, namely, by parsing out the contribution of a potential interaction and mediation between blast exposure and probable PTSD. In the fully adjusted model, the CDE, which estimates the effect between blast exposure and hearing difficulty, excluding the average effect of PTSD, was approximately 54% of the total association. The PIE, which estimates the effect due to mediation by probable PTSD and not interaction, was approximately 37% of the total association. There was no indication of either a reference or mediated interaction between blast exposure and probable PTSD. When the respective components are combined, interaction (reference interaction + mediated interaction) accounts for approximately 9% of the total association and appears negligible.

    Table 5. Percent decomposition results for the unadjusted and adjusted model.

    Effect Unadjusted
    Adjusteda
    % mediated 95% CI % mediated 95% CI
    Controlled direct 49% [12, 87%] 54% [18, 90%]
    Reference interaction 6% [−10, 22%] 4% [−16, 25%]
    Mediated interaction 10% [−15, 35%] 5% [−17, 27%]
    Pure indirect 35% [12, 58%] 37% [10, 63%]

    Note. Percent (%) mediated and 95% confidence interval (CI) are displayed.

    aAdjusted for gender, age, race/ethnicity, education, marital status, service branch, service component, service duration, and combat deployment.

    Discussion

    We examined blast exposure and self-reported hearing difficulties among post-9/11 military Service members and Veterans who have normal pure-tone hearing sensitivity, and we addressed the potential mediating role of PTSD in this population. We found that individuals with blast exposure had a higher prevalence of probable PTSD, as well as higher self-reported hearing difficulty than individuals who were not blast exposed. Our finding that blast exposure was associated with self-reported hearing difficulty among Service members and Veterans without peripheral hearing loss is generally consistent with the previous literature and what has been observed anecdotally in VA audiology clinics (Bressler et al., 2017; Gallun, Diedesch, et al., 2012; Gallun et al., 2016; Saunders et al., 2015).

    We also found that individuals with probable PTSD and normal pure-tone hearing sensitivity had a higher prevalence of self-reported hearing difficulty than individuals who do not have probable PTSD. Service members and Veterans with probable PTSD had a threefold higher prevalence of self-reported hearing difficulty compared to individuals without probable PTSD, after adjusting for potential confounders including blast. This is a unique contribution to the literature and suggests that PTSD may play an important role in the association between blast exposure and self-reported hearing difficulty. Previous observational studies have noted the high correlation between blast exposure and PTSD and generally conclude the association between blast exposure and hearing difficulty is complicated by the presence of PTSD (Gallun, Diedesch, et al., 2012; Gallun, Lewis, et al., 2012; Gallun et al., 2016; Saunders et al., 2015). Formal mediation analysis suggested that PTSD mediated a high proportion of the blast–self-reported hearing difficulty association (41%). This finding has important clinical implications, as it suggests that individuals with normal pure-tone hearing sensitivity who complain of hearing difficulties and have a positive history of blast exposure may benefit from an evaluation by mental health providers for untreated PTSD symptoms.

    We extended previous observational studies of blast exposure and hearing difficulty by estimating the mediating role of PTSD. We report a direct association between blast exposure and hearing difficulty, and an indirect association occurring through probable PTSD. There was little support for interactive effects within the mediation framework. These findings suggest a chain of risk, whereby one exposure contributes to subsequent exposures and each adverse experience raises the risk of the outcome (without the links on the chain interacting). These results support the conceptual framework that suggests that PTSD may lead to poor health outcomes through psychological and attentional mechanisms that alter symptom perception (Schnurr, 2015). Thus, individuals with PTSD are more likely to report poor health in the absence of disease/injury compared to individuals without PTSD (Schnurr, 2015). By understanding this risk process, interventions can be directed toward mediating factors as well as the primary exposure itself. This perspective may also be helpful when considering the complex health outcomes of war-related exposures.

    The presence of PTSD mediation does not rule out concomitant auditory processing difficulties among participants. Considering the magnitude of the observed mediation, our results suggest that the association between blast exposure and self-reported hearing difficulty among those with normal pure-tone hearing sensitivity may also be mediated through other pathways. From an audiologic perspective, possible mechanisms include subclinical damage to the cochlea (i.e., undetectable peripheral hearing loss; Kujawa & Liberman, 2009) and central auditory nervous system dysfunction (Gallun, Lewis, et al., 2012; Tepe et al., 2020). Some early evidence exists to support each of these mechanisms, although prospective studies capable of empirically testing such mediation are lacking. From a mental health perspective, other possible mechanisms and factors include depression, addiction, and non-PTSD-related anxiety. Each of these potential contributors warrants further investigation. Investigators wishing to further explore these mechanisms should utilize a causal framework to clearly identify hypothesized causal relationships. Explicit causal relationships coupled with clear causal questions will help to elucidate potential effects of blast exposure on auditory function (Fox et al., 2020).

    The main strength of this study lies in its use of an explicit causal framework based on theory and prior knowledge. We investigated each segment of the hypothesized pathway including assessment of mediation, which helps to advance discovery from association toward causality and mechanistic understanding. We also improve on past reports of the blast–hearing difficulty association by using primary collected data in a larger sample size. Limitations include the use of self-rating/reporting for the exposure, mediator, and outcome variables, use of a screening measure for determining probable PTSD, complete-case analysis, and the lack of a longitudinal data structure. Conclusions regarding causality are constrained because of the cross-sectional study design that precludes establishing temporality. Furthermore, even when using a causal mediation framework, it is unclear whether results estimated in existing data will translate into actual effects of potential future interventions (i.e., transportability is unknown; Westreich et al., 2017). Policymakers should be cautious in applying results of this and other studies to different, nonstudy contexts (including the same system at a different time); inverse probability of sampling weights could be used to assess transportability (Westreich et al., 2017).

    Notwithstanding these caveats, our findings have possible clinical implications both in the war theater and in audiology clinics. For individuals exposed to blasts, early interventions for stress reduction to mitigate the onset of PTSD may be an important focus. In the clinic, audiologists should consider screening for PTSD with appropriate referrals to mental health when warranted for Service members and Veterans with normal pure-tone hearing sensitivity who report hearing difficulty. Integrated treatment protocols that address both PTSD and hearing difficulty may also be helpful.

    In conclusion, we identified a greater proportion of Service members and Veterans with self-reported hearing difficulty among those with, versus without, PTSD symptoms and among those with, versus without, blast exposure while controlling for potential confounders. We further provide novel insight into the mechanistic nature of the associations between blast exposure and hearing difficulty. Formal causal mediation analysis suggests that some of the self-reported hearing difficulty in Service members and Veterans with blast exposure may be in part due to PTSD. Hearing care providers should consider the important role PTSD plays in fostering hearing difficulty.

    Disclaimer

    The views expressed in this manuscript are those of the authors and do not necessarily represent the official policy or position of the Defense Health Agency, Department of Defense, or any other U.S. government agency. This work was prepared as part of official duties as U.S. Government employees and, therefore, is defined as U.S. Government work under Title 17 U.S.C.§101. Per Title 17 U.S.C.§105, copyright protection is not available for any work of the U.S. Government.

    Supplemental Material on Figshare

    Acknowledgments

    This work was supported by a Department of Defense Congressionally Directed Medical Research Program Investigator-Initiated Research Award (PR121146), a Joint Warfighter Medical Research Program Award (JW160036), and a U.S. Department of Veterans Affairs (VA) Rehabilitation Research and Development (RR&D) Research Career Scientist Award (1 IK6 RX002990-01). This material is the result of work supported with resources and the use of facilities at the VA RR&D National Center for Rehabilitative Auditory Research (VA RR&D NCRAR Center Award; C9230C) at the VA Portland Health Care System in Portland, Oregon, as well as the Department of Defense, Hearing Center of Excellence in San Antonio, Texas. Portions of this article were presented at the American Auditory Society Annual Meeting, Scottsdale, AZ, March 5–7, 2020.

    References

    Author Notes

    Disclosure: The authors have declared that no competing financial or nonfinancial interests existed at the time of publication.

    Correspondence to Kelly M. Reavis:

    Editor-in-Chief: Ryan W. McCreery

    Editor: Ann Clock Eddins

    Additional Resources