Feasibility of Intensive Ecological Sampling of Tinnitus in Intervention Research

Abstract

Objective:

To assess whether adults with bothersome tinnitus will complete multiple ecological momentary assessments (EMA) per day, including during a therapeutic intervention trial. EMA is defined as repeated sampling of subjects’ current behaviors and experiences in real-time.

Study Design:

12-week longitudinal cohort study, conducted from August 28, 2017 to December 14, 2017.

Setting:

Online only.

Subjects:

Thirty adults with self-reported bothersome non-pulsatile tinnitus of >6 months duration.

Methods:

Participants completed 2 weeks of EMA text-surveys 7 times per day (pre-intervention), followed by 8 weeks of EMA questions 4 times per day (during-intervention), concluding with 2 weeks of EMA questions 7 times per day (post-intervention), for a total of 420 surveys over 12 weeks. During the 8-week intervention-period, participants completed a commercially available auditory-intensive online cognitive brain training program 20 minutes per day, 5 times per week (total of 800 minutes). The primary outcome measures were compliance with EMA surveys, as measured by survey response rates, and participant-reported effects of EMA on their tinnitus bother.

Results:

Of the 30 participants in this study (20 women and 10 men; median [range] age 54 [47–64] years), 25 participants completed the study protocol (83%). Participants completed a median of 87% of EMA surveys (range: 67–99%). Qualitative analysis of participants’ free-text responses found that participants did not report negative side effects of the EMA.

Conclusion:

Excellent participant compliance can be achieved with multi-week temporally rigorous EMA sampling. EMA sampling can successfully be conducted during an intervention. EMA is a promising sampling methodology in tinnitus research.

INTRODUCTION

Tinnitus is the perception of a sound that has no external source.¹ It is commonly described as “ringing or hissing” in the ear, in the absence of an acoustic stimulus. Tinnitus is a prevalent condition that affects approximately 40 million adults in the United States.² Across various nationalities, surveys estimate that the prevalence of adults with tinnitus is 10 to 25%.¹ Although many treatments are available for tinnitus, no cures for tinnitus currently exist.³ Some treatments for tinnitus aim to mask the sound, whereas other treatments, such as cognitive behavioral therapy, attempt to manage the secondary consequences of tinnitus.¹ Tinnitus is associated with significant functional and emotional impairment, including impaired sleep,⁴ anxiety,⁵ and depression.⁶ We use the phrase “tinnitus bother” to encompass the physical problems, functional limitations and emotional consequences of tinnitus.

Despite considerable focus on methods for reducing tinnitus bother, one of the primary limitations in the field of tinnitus research is the method by which tinnitus bother is measured. Standard measures of tinnitus bother, including the Tinnitus Functional Index (TFI)⁷ and Tinnitus Handicap Inventory (THI),⁸ use retrospective self-reporting measures. These retrospective indices are inherently prone to recall and recency bias.^9,10 Furthermore, fluctuation in tinnitus bother is a phenomenon described by many tinnitus sufferers; however these static retrospective indices cannot capture the dynamic and fluctuating nature of tinnitus bother.¹¹

An alternative to retrospective reporting measures is ecological momentary assessment (EMA).¹² EMA focuses on a participant’s “in the moment” experiences, thereby limiting recall bias and capturing moment-to-moment fluctuations.¹³ Pilot studies have shown that EMA is a feasible way to measure tinnitus bother, and participants had generally high levels of compliance.^11,14,15 However, these prior studies investigated the use of EMA either as pre-intervention assessment, or as an assessment tool not coupled to an intervention. No study to date has explored the impact of EMA assessment of tinnitus during an intervention to measure dynamic changes in tinnitus as a result of an intervention.

A major domain of tinnitus functional impairment for patients is cognitive difficulties.¹⁶ These cognitive difficulties have been substantiated in neurocognitive testing, where tinnitus patients have been shown to have poorer working memory,¹⁷ slower reaction times under dual-task conditions,¹⁸ and deficiencies in attention switching.¹⁹ Therefore, cognitive training programs, such as the commercially available program BrainHQ, seem to hold promise as a future therapy for tinnitus patients.²⁰ They are designed to use the brain’s capacity for neuroplasticity to improve functional connectivity in key networks associated with a variety of functions, including memory, perception, and cognition. No studies to date have examined a commercially available cognitive brain training program as an intervention for tinnitus patients.

Given these limitations in the current literature, the primary aim of our study was to use EMA over a full pre-/during-/post-intervention study period, to assess the viability of EMA as an intensive assessment of tinnitus bother over several months. Our second aim was to use EMA to assess the efficacy of a commercially available brain training program in reducing tinnitus impairment for adults with bothersome tinnitus. Elucidating the feasibility of EMA in tinnitus interventions carries important implications for tinnitus research study design and choice of sampling methodology.

METHODS

This study was conducted from August 28, 2017 to December 14, 2017. Participants were required to be between the ages of 20 and 70 years of age, and bothered by non-pulsatile tinnitus for more than 6 months. The upper age limit of 70 was set because we expected that there would be a reduced capacity for neuroplasticity in older age groups, and we were studying an intervention based on neuroplastic mechanisms of action. Level of bother was assessed using the Overall Global Rating of Bother Scale (GBS), a 1-item self-report questionnaire that has been modified from the Global Impressions Scale²¹ to assess tinnitus experience. Using the GBS, candidates reporting to be “Bothered more than a little but not a lot,” “Bothered a lot,” or “Extremely bothered” were included. Candidates answering “yes” to the question, “Does your tinnitus typically sound like it is pulsing, such as a heart-beat in your ear?” were excluded, given that we were interested in studying non-pulsatile tinnitus.

Participants were required to have access to a smartphone device with texting and data, as well access to a computer for the duration of the study. Candidates with hearing loss not addressable with hearing aids and candidates who could not use headphones were excluded, based on answering “no” to the questions, “Is your hearing loss corrected to normal with the use of hearing aids?” and/or “Can you use headphones?” Participants with depression, as measured by a score of 10 or greater on the PHQ-9²², which reflects moderate depression severity or greater were excluded because of concern for the confounding effects of their mental illness, informed of their score, provided with counseling resources, and encouraged to seek care from their primary care physician. All participants were required to read, write, and understand English. Additionally, candidates with any unstable psychiatric conditions were excluded. “Unstable psychiatric condition” was defined as not yet achieving a stable dose of medication, or less than 1 month on their most recent psychiatric medication.

The Washington University Otolaryngology Research Participant Registry was queried to identify potential participants with tinnitus. All matching registry members were sent an email with a description of the study and a link to an online screening survey. This online screening survey was emailed to candidates using the RedCap version 7.3.5 survey tool (Vanderbilt University). If a potential participant met eligibility requirements, a text message was sent to their mobile phone in order to assess their ability to send/receive messages. Participants passing both phases of screening were offered study participation. A written informed consent document was provided to all eligible participants via email, and eligible participants were given the opportunity to ask questions via phone or email. This study was approved by the Washington University Institutional Review Board prior to study initiation.

Consented participants completed baseline assessments, which included an assessment of tinnitus severity using the Tinnitus Functional Index (TFI);⁷ a measure of anxiety using the Beck Anxiety Inventory (BAI);²³ as well as the Oregon Hearing Research Center forms²⁴ for Medical and Health Information, Hearing History and Occupation Exposure, and Tinnitus Description. Baseline assessments were administered as electronic forms using the RedCap version 7.3.5 survey tool (Vanderbilt University).

Following the baseline assessments, participants received text message surveys between the hours of 8:00 am and 8:00 pm (scheduled based on their time-zone) for 12 weeks (420 texts in total). For the first 2 weeks (pre-intervention), the text surveys were delivered every 2 hours (7 texts per day). For the next 8 weeks (during-intervention) text surveys were delivered every 4 hours (4 texts per day). For the last 2 weeks (post-intervention) text surveys were delivered every 2 hours (7 texts per day). Text frequency was decreased during the study intervention to decrease participant burden. This EMA protocol was established based on expert consultation and our lab’s prior experience with EMA. Each text contained a link to an online survey assessing current tinnitus bother and various measures of anxiety and depression (For a detailed overview of survey questions, see Table 1). Participants responded to the survey items using 0–100 sliding scales. All texts were delivered using the EzTexting (Santa Monica, CA) electronic text delivery system. All text surveys were administered using the Qualtrics version 8.2017 survey platform (Provo, UT).

Table 1.

EMA questions

1x Daily Today…	Start of Day	1. Upon awakening, how rested do you feel?
	End of Day	2. How much trouble did you have staying awake while engaging in social activities?
		3. How difficult was it for you to find the right word while talking, texting or emailing?
At Each Time Point Right now…		1. How loud is your tinnitus?
		2. How bothered are you by your tinnitus?*
		3. How anxious or stressed do you feel?
		4. How difficult is it for you to concentrate on what you are doing?‡
		5. How difficult is it for you to avoid thinking about your tinnitus?
		6. How difficult is it for you to follow a conversation?
		7. How difficult is it for you to focus?†
		8. How interested are you in activities or people you normally enjoy?
		9. How worried do you feel?
		10. How overwhelmed do you feel?

^*Related to bother

^†Related to focus

^‡Related to concentration

After 2 weeks of text surveys, participants were instructed to complete auditory intensive online brain training exercises using the Posit Science BrainHQ platform (San Francisco, CA). Participants were expected to use the program for 20 minutes per day, 5 days per week for 8 weeks (for a total of 800 minutes). Expectations for program use were established based upon consultation with Posit Science. Each participant was given an online account, and instructions to complete exercises in the “Personalized Training” Portal. The “Personalized Training” exercises were selected by the study team in consultation with scientists at Posit Science, because the exercises specifically targeted auditory brain processes. More information about the cognitive sub-domains targeted in each exercise is provided in Table 2. The study coordinator assessed participant compliance using the online BrainHQ study portal every 3 days and sent reminders to participants who were not meeting protocol requirements. Participants continued to answer text surveys during this 8-week period.

Table 2.

Brain HQ Exercises

Cognitive Domain	Brain HQ Exercise
Attention	Divided Attention
	Mixed Signals
Brain Speed	Fine Tuning
	Sound Sweeps
Memory	Hear, Hear
	Memory Grid
	Syllable Stacks
	To-Do List Training
People Skills	In the Know

After the 12-week study period, participants completed the TFI, BAI, and Patient Global Impression of Change (PGIC)²¹ assessments using the RedCap platform. A reduction in the TFI of 13 points or more for an individual has been established as clinically meaningful change.⁷ Additionally, participants were asked to respond using free-text to questions about their perceived benefit from the intervention and their feelings about the use of EMA. The free-text questions were: “Overall, how would you describe the response of your tinnitus to the brain training program (BrainHQ)?” and “Have you had any other changes in response to the brain training program (BrainHQ)? For example, changes in concentration or memory?”

Participant responses to these free-text questions were initially analyzed by three independent raters. Raters were undergraduate research assistants (Ras) who were unaware of participant scores or the overall scope of the study. They were informed only that participants had provided free-text responses regarding questions about their experiences during a study of tinnitus during which there was (a) EMA and (b) an intervention. Raters were asked to assess on a scale of 1 to 7 the following questions: (1) Does the person think their tinnitus has gotten better or worse? (Scale: 1=Much Worse, 7=Much Improved); (2) To what extent does the person report the training causing negative side effects? (Scale: 1=Not At All, 7= Very Much So); and (3) To what extent does the person report the EMA causing negative side effects? (Scale: 1=Not At All, 7= Very Much So). The third rater experienced some difficulty using the rating scale. Due to the lack of range in her responses using the rating scale, her ratings were not used in this assessment.

The primary outcome measures were compliance with EMA surveys, as measured by survey response rates, and participant-reported effects of EMA on their tinnitus bother. Secondary outcome measures were change in TFI, BAI, and PGIC, and compliance with BrainHQ.

Statistical analysis

After the conclusion of the study, standard descriptive statistics were used to explore distribution of demographic and tinnitus characteristics, as well as the distribution of the responses of EMA questions in any given study period. Normal distribution assumption was explored for all continuous level variables using histograms and Kolmogorov-Smirnov test. Friedman’s ANOVA was used to explore changes in the area under the curve (AUC) of the EMA responses for questions addressing bother, focus, and concentration. The area under the curve is equivalent to the average score of all EMA responses during each assessment time. Spearman’s correlation was used to explore correlation of the changes in pre- and post-intervention AUC with changes in participant TFI score.

Reliability of the raters was assessed using two-way random intraclass correlations (ICCs) for consistency of the average. Thus, the reliability values obtained indicate how well the raters would agree on the rank-ordering of the participants in terms of the codes. Finally, to assess whether EMA increased tinnitus bother prior to the intervention, we used a multilevel model to test whether tinnitus bother increased over the days of EMA prior to the intervention.

RESULTS

A total of 441 individuals on the Otolaryngology Participant Registry were sent an advertisement for the study, and 98 of these individuals completed the screening assessments. Of the 98 individuals, 39 met all inclusion and no exclusion criteria, and thus were offered participation in the study. 30 participants enrolled in this study. 5 participants withdrew their participation over the course of the study. Further details are included in Figure 1.

Figure 1.

Study design

Study Population

The median age of participants was 54 years old, with range from 31 to 70 years. Additional information about the study participants can be found in Table 3. A description of tinnitus characteristics is provided in Table 4. The median duration of tinnitus was 5 years (range: 0.5 to 35 years) and the median loudness of tinnitus measured on a scale 0–10 was 5 (range: 3 to 10). The majority of participants (n=16; 64%) reported to be bothered “more than a little, but not a lot” by their tinnitus. Many (n=13; 52%) had previously sought medical help for their tinnitus.

Table 3.

Description of characteristics of study participants (n=25)

	n (%)
Gender
Male	6 (24)
Female	19 (76)
Age
Median (range), years	54 (31–70)
Education
Associate Degree or some college	3 (12)
Bachelor Degree	9 (36)
Masters Degree or equivalent	8 (32)
Doctorate Degree or higher	5 (20)
Current Employment Status
Full Time	14 (56)
Part Time	3 (12)
Unemployed	1 (4)
Retired	7 (28)

Table 4:

Description of tinnitus characteristics for study participants (n=25)

	n (%)
Where does your predominant tinnitus sound seem to be coming from?
Ear	20 (80)
Head	4 (16)
Unsure	1 (4)
Tinnitus duration, median (range), years	5 (0.5–35)
Does your tinnitus interfere with sleep?
Yes, often	2 (8)
Yes, sometimes	13 (52)
No	10 (40)
What is the loudness of your usual tinnitus?
Median (range)	5 (3–10)
How much effort does it take to ignore your tinnitus when it is present?
Can ignore it with slight effort	8 (30)
Can ignore it with some effort	10 (40)
Can ignore it with considerable effort	6 (24)
Can never ignore it	1 (4)
How much discomfort do you usually experience from your tinnitus?
No discomfort	1 (4)
Very mild discomfort	8 (32)
Mild discomfort	9 (36)
Moderate discomfort	7 (28)
Have you previously sought medical help for your tinnitus?
Yes	13 (52)
No	12 (48)
Difficulties hearing speech
Yes, often	5 (20)
Yes, sometimes	14 (56)
No	6 (24)
Difficulties with other types of sounds
Yes, often	4 (16)
Yes, sometimes	12 (48)
No	9 (36)
Ever worn a hearing aid
Yes	5 (20)
No	20 (80)

Compliance with EMA

Participants completed a median of 87% of text surveys (range: 67% to 99%) that were sent to them, equating to a median completion of 365 surveys per participant, over the study period.

Quantitative Coding of Free Text Responses

Analysis of the reliability of the raters showed that the raters had good reliability for all three questions, with Two-Way Random ICCs for consistency > .78. Examining the average ratings for these items, the coders indicated that most participants (n = 18) seemed to have experienced no change in response to the intervention (coder ratings of 3.5 to 4.5 for those participants, where 4 is no change), with only two participants showing signs of improvement. However, most participants did not report negative side effects of the training (n = 18 had an average rating of 1 or 1.5 on item 2, where 1 is not at all in terms of report of negative side effects). Participants also did not report negative side effects of EMA overall (n = 20 had an average rating of 1 or 1.5 on item 3). Of those reporting negative side effects, only 2 participants for the intervention and 4 participants for EMA had average ratings approaching or reaching Moderately (i.e., average ratings of 3.5 to 5). No average ratings exceeded 5 (where 7 is very much so) for either side-effect rating.

Change in TFI, PHQ-9, and BAI

The median Tinnitus Functional Index score at baseline was 33.2 points (range: 10–88), the median PHQ-9 was 3 points (range: 0–9), and median BAI scores was 5 points (range: 0–45). The median change between pre- and post-intervention for TFI score was (−2.4 (95% CI: −7.4 to 2.8) and for BAI was: −1.0 (95% CI: −3.5 to 0.5).

EMA Domains

Change in the EMA domains of focus and concentration were moderately to strongly correlated with change in TFI (r=0.462 and r=0.592 respectively). The change in EMA Bother was weakly correlated with change in TFI (r=0.305).

Test for Reactivity to EMA

Across the sample of participants completing any part of the pre-intervention EMA (n = 27), level of tinnitus bother was not predicted by day of study (p = .219). There was thus no evidence of a systematic increase in tinnitus bother across initial EMA days.

PGIC

As shown in Figure 2, the majority of participants reported “no change (or condition has gotten worse)” or ”almost the same, hardly any change at all” in response to the intervention within the domains of activity limitations (76%, n=19), symptoms (88%, n=22), emotions (76%, n=22), and overall quality of life (84%, n=21) on the PGIC assessment.

Figure 2:

Change in domains related to tinnitus on the PGIC scale

BrainHQ Compliance

Participants spent a median of 920 minutes (range: 178 to 1438 minutes) in the BrainHQ program over the duration of the study. The majority (n=16; 64%) of participants met the time requirement for BrainHQ use.

DISCUSSION

In this study, participant compliance in completing EMA text surveys was excellent; with a median compliance rate of 87%. Additionally, few participants reported negative side effects of EMA, and those who did, did not strongly report such effects. In response to the 8-week cognitive brain training intervention, there was no statistically or clinically significant change in tinnitus severity, anxiety, or depression.

This is the first study to use EMA as a sampling methodology during a tinnitus intervention. Our experience suggests that EMA can be used successfully in this setting, as evident by our high rate of compliance with EMA and low rate of trial withdrawals attributed to bother from EMA sampling (n=2 withdrew due to “too many surveys”). EMA carries exciting prospect as a research tool, particularly given its potential to minimize biases inherent to retrospective reporting measures,²⁵ its ability to provide researchers with insight into longitudinal intra- and inter-participant variability,¹¹ and its ability to track participants’ experience across various phases of a study.²⁶

Our finding that EMA can be used without negative side effects fits within the context of other EMA findings for behavioral health and psychiatric disorders. McCarthy et al studied the use of high frequency (6x/day) vs low frequency (1x/day) EMA in adults trying to quit smoking.²⁷ They found that EMA frequency was unrelated to participants’ smoking cessation or prolonged abstinence. Additionally, Pfalz et al found that the use of EMA 3x/day during a study of patients with panic disorder and PTSD did not disturb either group.²⁸ That said, there are some indications in the literature that a higher frequency of sampling during the day might slightly increase participant burden, with EMA sampling closer to 3x/day potentially conferring minimal burden.²⁹ To our knowledge, our study represents the longest and most temporally rigorous use of EMA sampling to date, further supplementing existing evidence that EMA is a tolerable sampling methodology for study participants. However, we would consider lowering the daily frequency of our EMA protocol in future studies if it is possible to both achieve statistical aims and obtain minimal burden by doing so.

Furthermore, some studies have shown potential patient outcome benefits of EMA in psychiatry.^27,30 Dewey et al showed that EMA 6x/day for 2 weeks in PTSD patients significantly reduced symptom severity.³¹ They hypothesized that EMA may serve as an exposure to facilitate emotional processing. This hypothesis is consistent with several psychotherapeutic approaches, including Mindfulness Based Stress Reduction (MBSR) for tinnitus. The central tenet of MBSR is increased awareness of the present moment (even if that moment includes tinnitus). The success of MBSR in several tinnitus studies,^32,33 as well as the tolerability of EMA, both support the foundational notion that increased awareness does not lead to increased distress—at least not for participants in general. This is an important principle for researchers considering the use of EMA in research study design, since this principle would suggest that sampling participants repeatedly does not lead to increased distress.

The lack of observed effect in participants’ response to the cognitive brain training program stands in contrast with prior work in various disease processes.^34–38 Specifically within the field of tinnitus, Kallogjeri et al. found that cognitive training for tinnitus was associated with self-reported improvement in attention, memory, and perception of tinnitus.²⁰ There are several explanations for our lack of observed response. The first could be that our outcome measures may have failed to capture the intervention effect. Second, this was a small study, so both Type 1 and Type 2 errors are possible. Third, it could be that cognitive training does not work for tinnitus, or that cognitive training could work, but BrainHQ training does not. Lastly, it could be that our study design with the use of EMA interfered with our intervention. However, we believe this last explanation to be unlikely, given our above findings. Our study’s lack of a control group precludes us from drawing definitive conclusions about the treatment efficacy of BrainHQ. Future work to further explore the efficacy of cognitive brain training in tinnitus is needed.

The limitations of this study include the study population, which consisted of mostly white, highly-educated subjects, who owned a smart phone, and were either employed or retired, with a predominance of women participants. The results from this study therefore may not be generalizable to a broader population. Without a control group, it remains possible that the null effects for the intervention might somehow be due to the use of EMA. However, the fact that tinnitus bother did not increase across the pre-intervention assessment argues against this possibility. Additionally, we may have failed to see an effect of BrainHQ because certain individuals may experience tinnitus bother more in the domains of sleep or emotions, which wouldn’t necessarily be affected directly by cognitive brain training. We would encourage future investigators studying cognitive brain training to further refine their selection criteria to more intentionally capture patients who experience cognitive disturbances as a manifestation of their tinnitus.