Patients with subjective tinnitus were enrolled from the tinnitus specialist clinic of Eye & ENT Hospital of Fudan University from July 14, 2020, to May 24, 2021. All the included participants were Mandarin speakers, aged over 18 years, and able to communicate with instructors without any hearing aids. The average pure tone threshold (0.5, 1, 2, 4 kHz) should be no more than 55 dB HL of the worse ear. The participants were required to understand the purpose and process of the study, complete at least 3 follow-up visits in the clinic, and sign the informed consent prior to the commencement of the experiment. Patients with pulsatile tinnitus, objective tinnitus, or clear primary causes such as otosclerosis, Meniere disease, and acute stage of sudden sensorineural hearing loss were excluded from this study. Moreover, for safety reasons, patients with severe health issues such as cardiovascular and cerebrovascular diseases or tumors were excluded from this study. This was a nonrandomized concurrent controlled trial, and we followed the TREND (Transparent Reporting of Evaluations with Nonrandomized Designs) guidelines for nonrandomized controlled trials (Multimedia Appendix 1). A total of 107 eligible patients who tended to use the FTRS app for acoustic therapy received DFCRS, and they were labeled as the experimental group. The other 77 patients who refused to use the app received UM, serving as the control group. The experimental group and the control group were recruited and provided acoustic therapy at the same time. All participants completed 4-timepoint follow-up questionnaires.

All tests were approved by the institutional review board of Eye & ENT Hospital of Fudan University (2017048). All patients signed the informed consent forms, and sensitive information relevant to patient privacy was removed before the analysis. Both forms of acoustic therapy did not add any cost to the patient, and each patient could also receive an appointment or a plus number service for the tinnitus specialist clinic.

The baseline data were collected mainly from 4 aspects: general information, characteristics of tinnitus, audiological and tinnitus assessments, and results of the multidimensional assessment scales of tinnitus. General information included details on age, gender, medical history, and treatment history. The characteristics of tinnitus included information on position, course, pitch, and correlative factors of tinnitus. Audiological and tinnitus tests included pure-tone audiometry (125 Hz-8 kHz), acoustic immittance, and tinnitus pitch matching. Five standardized questionnaires in the Chinese version were used to evaluate the severity of tinnitus and the QoL of patients with tinnitus from multiple dimensions.

In the DFCRS-treated group, according to precise audiological and tinnitus tests, a series of 4 endpoints of the modulation range adjacent to the tinnitus pitch were calculated using our specific algorithm (Multimedia Appendix 2). Based on these 4 endpoints, 2 specific frequency bands of 1/3 octave intervals below and above the tinnitus pitch were confirmed. Thereafter, with the assistance of the FTRS app, soft and relieving music was processed by dynamically increasing the intensity of both specified bands with a 10-dB gain, thus outputting the personalized relieving sound for tinnitus. Accordingly, in the UM-treated group, the unmodified original music was used.

After baseline data completion at the initial point in time (T0), patients in the DFCRS-treated group were trained to use the FTRS app correctly, while others in the UM-treated group were instructed to listen to UM through their favorite players. All participants were required to listen to the modified or unmodified sound in a quiet environment with a comfortable volume to avoid possible hearing damage.

In this study, the participants were not allowed to receive any other simultaneous treatments that might interfere with sound therapy. To obtain a better effect, all patients were encouraged to be exposed for as much time as possible in the therapeutic sound environment, and the recommended duration was at least 2 h/day. In addition, the participants were instructed to record the time of daily sound exposure accurately and truthfully. Finally, researchers maintained contact with the participants through various accessible means such as phone calls, WeChat, emails, and tinnitus specialist clinic visits.

Well-trained otologists implemented counseling and education for all patients with tinnitus. The main topics included the etiology and basic pathogenesis of tinnitus, the relationship between tinnitus and hearing loss, the interpretation of audiological and tinnitus test results, the psychoacoustic characteristics of tinnitus, the relationship between tinnitus and attention, guidance on exploring common methods to divert attention, and relaxation training. During the entire procedure, the participants’ confusion and discomfort with regard to tinnitus were patiently, promptly, and sufficiently dispelled.

There were 4 timepoints in total, including the baseline when the participants were enrolled and at 1, 2, and 3 months after commencing treatment. The primary outcome was the THI scores from baseline to follow-up endpoints. In addition, the other assessment instruments (HADS, AIS, TCS, and FTQ) as secondary outcomes also provided longitudinal insights into the efficacy of tinnitus treatment.

Statistical analyses were performed using R software (version 4.1.2; R Foundation for Statistical Computing) with the lme4, nnet, and ggpubr packages. Two-sided values of P<.05 were deemed statistically significant. Demographics and baseline characteristics such as age, sex, tinnitus course, location, tinnitus pitch, severity, and assessment instruments (THI, HADS-A, HADS-D, AIS, FTQ, and TCS) were summarized by groups that were defined by the treatment results. We summarized the descriptive statistics of the continuous variables (mean, standard deviation or median, and quartile) and categorical variables (number and proportion of participants in each category). Differences in the characteristics among the groups were tested with chi-square for categorical variables; 1-way analysis of variance was used for normally distributed continuous variables, and the Kruskal–Wallis test was used for nonnormal continuous variables. A 2-tailed t test was applied to compare the mean multidimensional assessment data before and after treatment and to determine the differences between subgroups. Linear mixed models were used to test for changes in the THI scores across 3 months of acoustic treatment between groups (DFCRS or UM treatment) and time. A multiclass logistic model was built with a stepwise function to determine the influence of different covariates on the effect of acoustic treatment, and Akaike information criterion/Bayesian information criterion/corrected Akaike information criterion was used to select the best model.

A total of 184 patients with tinnitus (107 in the DFCRS treatment group and 77 in the UM control group) completed the baseline data collection and 3 follow-up visits. After the 3-month treatment, patients were asked to report the effectiveness of tinnitus treatment based on their subjective perceptions. The baseline data are shown in Table 1. In the DFCRS treatment group, 50.5% (54/107) were men and 49.5% (53/107) were women, aged 45 (range 35.5-57) years, with a tinnitus duration averaged at 12 (range 4-24) months. In the UM group, 55% (42/77) were males, the age was 50 (range 37-59) years, and the tinnitus duration was 8 (range 3-42) months. The patients’ gender and age did not differ significantly between the experimental group (DFCRS) and the control group (UM). We also compared other baseline data and observed no significant differences in the distribution of tinnitus characteristics involving location, frequency, and severity assessed by multidimensional tinnitus scales (HADS, AIS, Visual Analog Scale, FTQ, and TCS) between the 2 groups. Besides, there were no significant differences in the hearing levels in either ears between the 2 groups. However, the THI scores differed at baseline between the 2 groups (UM group: median 40, range 24-60; DFCRS group: median 48, range 33-66; P=.01). The overall baseline data were comparable between the DFCRS group and the UM group, suggesting that the following therapeutic effect was not significantly influenced by pre-experimental demographic information and characteristics of tinnitus.

As the primary outcome, the THI scores were calculated using the linear mixed model in both DFCRS and UM treatment groups. The results of the linear mixed model revealed significant main effects of follow-up time (estimate=–0.119; P<.001) (Table 2), which showed a general decrease in the THI scores over the trial duration. Besides, there were significant effects in the DFCRS group compared to those in the control group, and the negative value revealed a remarkable treatment effectiveness of DFCRS measured by THI compared to that of UM in the control group (estimate=–16.6469; P<.001).

Apart from the follow-up time and treatment procedure, no other factors were found to contribute to a significant effect in the linear mixed model. Comparisons of the multidimensional assessment scales between the DFCRS-treated group and the UM control group at 4 timepoints are shown in Figure 1. During the 3 months of treatment, almost all of the scores on the 6 scales in the DFCRS-treated group showed significant decline compared to those in the UM control group, indicating better effectiveness of DFCRS in managing tinnitus. We found that UM had no significant therapeutic effect on tinnitus (Table S1 in Multimedia Appendix 3). Moreover, the differences were mostly maintained or were even enlarged across time especially after 1 month of treatment. No adverse effects were reported during follow-up in both groups.

To further identify the factors affecting the effectiveness of DFCRS, we analyzed the statistical relationships among gender, age, acoustic treatment time per day, various tinnitus characteristics, and treatment results according to patients’ subjective reports. After 3 months of DFCRS treatment, 86.9% (93/107) of the patients reported tinnitus relief, 5.6% (6/107) reported that tinnitus disappeared, and 7.5% (8/107) reported worsened tinnitus (Table S2 in Multimedia Appendix 3). The patients’ gender and age did not differ significantly among the 3 different treatment outcome groups. There were no significant differences in the distribution of characteristic tinnitus indicators such as location, tone, course, and severity assessed by multidimensional tinnitus scales among the 3 subgroups as well. Interestingly, the data showed that there was a significant difference in the treatment time per day among the 3 groups (P=.007). Participants whose tinnitus was alleviated or disappeared received sound therapy for a longer time than those who reported worsening tinnitus, indicating a time-dose–dependent relationship for implementing DFCRS as a tinnitus treatment strategy (Table S2 in Multimedia Appendix 3). We then conducted a two-by-two comparison among the 3 subgroups, and the results revealed that significant differences existed between the worsening tinnitus group and the tinnitus relief group (P<.01) as well as the worsening tinnitus group and the tinnitus disappeared group (P<.05; Figure 2). This finding preliminarily confirmed that an increase in the exposure of DFCRS treatment contributed to better efficacy in managing tinnitus.

For more precise analysis, we built a multiclass logistic model to determine the influence of different covariates on acoustic treatment effects (Table 3). The results showed that compared to patients whose tinnitus worsened, patients who reported tinnitus relief had a longer treatment time per day (odds ratio [OR] 13.07, 95% CI 4.19-40.78; P<.001), and patients who reported tinnitus disappearance had the longest daily treatment time (OR 64.78, 95% CI 11.63-360.77; P<.001). In addition, patients with more severe tinnitus at baseline tended to respond better to acoustic treatment. More specifically, patients with severity III were more likely to have tinnitus relief (OR 40.37, 95% CI 3.42-477.14; P<.001) than patients whose tinnitus severity level was I. Similarly, patients with severity IV were more likely to have tinnitus relief (OR 10.46, 95% CI 2.12-51.62; P<.001) or tinnitus disappearance (OR 83.71, 95% CI 5.9-1188.28; P<.001) than those with tinnitus severity I.

After 3 months of DFCRS treatment, the scores of all the multidimensional scales significantly decreased compared with the baseline data (P<.001 for all scales), indicating the efficacy of DFCRS therapy in managing tinnitus. The THI score decreased from 48.00 (95% CI 32.00-66.00) points at baseline to 30.00 (95% CI 18.00-42.00) points at the 1-month follow-up, to 22.00 (95% CI 12.00-36.00) points at the 2-month follow-up, and finally to 18.00 (95% CI 8.00-30.00) points at the 3-month follow-up. Significant differences were observed between the scores at baseline and 3 consecutive months of follow-up (P<.001; Figure 3).

By comparing the 3 follow-up scores with the baseline of the other secondary outcomes, we observed that a more significant decline occurred in the following follow-up than in the preceding one (HADS-A, AIS, FTQ: P<.001 at 1-month follow-up, P<.001 at 2- and 3-month follow-ups; and HADS-D, TCS: P<.01 at 1-month follow-up, P<.001 at 2- and 3-month follow-ups). These significant differences suggested that there was a time-dependent increase in the effectiveness during the experimental procedure.

We also analyzed the differences between 2 of the 3 follow-up outcomes of the 6 assessment scales. Regarding THI scores, significant differences were observed between 2 and 3 months of follow-up (P<.05) and between 1 month and 3 months of follow-up (P<.001). In contrast, the THI scores did not decrease significantly from 1 month of follow-up to 2 months of follow-up (P=.07). A similar reduction was also found in the TCS scores (1 month vs 2 months: P=.09; 1 month vs 3 months: P<.01; 2 months vs 3 months: P=.17; baseline vs 3 months: P<.001), indicating that the therapeutic effect of DFCRS on tinnitus impairment and catastrophizing was observed at 2 months and was maintained in the third month.