One purported use of low-level laser therapy (LLLT) is to promote healing in damaged cells. The effects of LLLT on hearing loss and tinnitus have received some study, but results have been equivocal. The purpose of this study was to determine if LLLT improved hearing, speech understanding, and/or cochlear function in adults with hearing loss. Using a randomized, double-blind, placebo-controlled design, subjects were assigned to a treatment, placebo, or control group. The treatment group was given LLLT, which consisted of shining low-level lasers onto the outer ear, head, and neck. Each laser treatment lasted approximately five minutes. Three treatments were applied within the course of one week. A battery of auditory tests was administered immediately before the first treatment and immediately after the third treatment. The battery consisted of pure-tone audiometry, the Connected Speech Test, and transient-evoked otoacoustic emissions. Data were analyzed by comparing pre- and posttest results. No statistically significant differences were found between groups for any of the auditory tests. Additionally, no clinically significant differences were found in any individual subjects. This trial is registered with ClinicalTrials.gov (NCT01820416).
Low-level laser therapy (LLLT) has been practiced for over 20 years in Europe and has more recently been introduced in the United States as a treatment for pain and postsurgical tissue repair. It has been proposed that laser energy in the red and near-infrared light spectrum may aid in the repair of tissue damage. A proposed mechanism for this therapeutic effect is the stimulation of mitochondria in the cells to produce more energy through the production of adenosine triphosphate [
It has been postulated that LLLT may improve cochlear function. Animal studies have found that laser stimulation can induce anatomic and physiologic changes in the cochlea. Rhee et al. [
Studies in humans have investigated the effects of LLLT on both hearing loss and tinnitus, with equivocal results. Some studies have found an improvement in hearing thresholds and tinnitus symptoms (e.g., [
Further research on the effect of LLLT on hearing in humans appears warranted. Although some studies showed improvements in hearing thresholds, no published study to date has examined the effect of LLLT on speech understanding and only one has examined the effect on cochlear function via otoacoustic emissions [
In order to accurately detect possible changes in hearing status due to laser treatment, it was necessary to avoid using subjects whose hearing might fluctuate due to other factors. Before potential subjects were enrolled in the study, they were asked a list of screening questions to determine eligibility. The questions were chosen to ensure stable hearing, as well as to address possible safety issues. All subjects were also required to have normal middle ear function, as assessed by 226 Hz tympanograms.
A total of 35 adult subjects were enrolled in the study. Two subjects withdrew from the study due to loss of interest and/or scheduling difficulty. The data from three additional subjects were not included in the analysis. One subject yielded unreliable audiometric and speech understanding data, speech scores could not be obtained from one subject with a profound hearing loss, and calibration problems compromised data from the third subject. Data from the remaining 30 subjects were included in the analyses. The experimental protocol was approved by the Institutional Review Board of The University of Iowa, and written informed consent was obtained from all participants.
An Erchonia EHL laser (Erchonia Medical, Inc.) was used to provide the laser stimulation. The device was a portable
The study used three groups: treatment, placebo, and control. Subjects were pseudorandomly assigned to one of the three groups. Initial group assignment was random with occasional adjustment to ensure that the three groups were similar in terms of number of participants, female/male ratio, mean age of participants, and mean pure-tone audiometric thresholds. The composition of each group is shown in Table
Group characteristics.
Group |
|
F/M | Age | PTA |
---|---|---|---|---|
Treatment | 9 | 0.44 | 52.3 (14.6) | 37.4 (15.9) |
Placebo | 10 | 0.30 | 58.9 (7.4) | 41.4 (10.3) |
Control | 11 | 0.45 | 47.1 (16.7) | 39.8 (18.6) |
| ||||
Total | 30 | 0.40 | 52.8 (12.9) | 39.6 (15.3) |
N: number of subjects in group. F/M: female to male ratio. Age: mean age in years. PTA: mean audiometric pure tone average (0.5, 1, and 2 kHz) in dB HL. Numbers in parentheses are standard deviations.
The treatment group received the laser treatment protocol (described in Section
The pretest, treatment sessions, and posttest took place in three scheduled visits within a 7–10-day period for each subject. On the first visit, the pretests were administered, followed immediately by the treatment session. Subjects in the treatment group received laser treatment with the functioning laser device. Subjects in the placebo group received laser treatment with the nonfunctioning device. Subjects in the control group simply sat in a comfortable chair and had a short conversation with the researcher. All subjects returned for a second treatment session 2-3 days after the first visit. Subjects returned for the third and final treatment 2-3 days after the second treatment session. Immediately after the third treatment, the posttest battery was administered.
Subjects did not know whether they were in the treatment or placebo groups, and the researchers administering the laser treatments did not know whether they were using the treatment or placebo device. Appropriate laser safety goggles (Laservision, style F12, filter 000131.000) were worn by subjects and the researchers administering the laser treatment. The goggles had lenses rated OD 6+ @ 510–680 nm, which blocked all visible laser light. The goggles also hugged the face firmly, preventing laser light from entering from the side.
To further avoid potential bias in the test results, the research team was divided into two groups. Two team members administered the laser therapy independently of two other team members that administered the battery of auditory tests. The members who administered the auditory tests also performed scheduling and group assignments. Groups were simply identified by colors; so, the testers did not know which treatment (laser, placebo, or control) any subject was receiving. The testers were not present during treatment sessions. The team members giving the laser therapy did not know which laser device was functioning and which was the placebo. A fifth member of the research team, not otherwise involved in the study, assigned the functioning and placebo laser devices to two of the color groups and was responsible for checking the devices weekly to ensure that the functioning laser device was working.
The LLLT treatment protocol was based on a pilot study conducted by HearingMed (unpublished) showing improvement of word recognition scores following LLLT. Subjects in the treatment group had the low-level laser applied for approximately 4 minutes to the area around both pinnae, the back of the neck, and the top of the head. Subjects in the placebo group received the same protocol, except that the disabled laser device was used. Subjects in the control group simply sat in a comfortable chair and conversed with the research team member for a few minutes, and no treatment of any kind was administered. The laser was applied as described in the following steps and as shown in Figure
Visual depiction of the laser treatment. Each step is described in detail in the text. The white circle represents the subject’s head. The double black lines represent the laser beams on the subject’s head. The white arrows show the directional movement of the laser beams (horizontal, vertical, or rotational).
The laser was centered on the right temporomandibular joint, just anterior to the external auditory meatus of the ear, at a distance of approximately 2 inches from the surface of the skin. The hand-held probe was rotated from vertical to horizontal and back continuously for 15 seconds.
The laser was positioned on midline of cervical spine with the beams running vertically from external occipital protuberance to the seventh cervical vertebrae. The hand-held probe was held at a distance of approximately 3 inches from the surface of the skin and continuously swept horizontally back and forth for 30 seconds.
The left temporomandibular joint was stimulated, as described in Step
The laser was positioned on top of the head with the beams running across the head from ear to ear. The probe was held at a distance of approximately 2 inches from the surface of the head and continuously swept back and forth from the forehead to the occipital protuberance for 30 seconds.
The laser was centered on the right external auditory meatus, with the probe held at a distance of approximately 2 inches from the surface of the pinna. The probe was rotated from vertical to horizontal and back continuously for 60 seconds.
The laser was positioned over the cervical spine with the beams running horizontally. The probe was held at a distance of approximately 2 inches from the surface of the skin and continuously swept up and down from the occipital protuberance to the top of the shoulders for 15 seconds.
The left external auditory meatus was stimulated, as described in Step
The auditory test battery consisted of three assessments: pure-tone audiometry, speech understanding, and transient-evoked otoacoustic emissions (TEOAEs). These tests were chosen to examine different aspects of hearing; pure-tone audiometry assessed auditory sensitivity in quiet, speech testing assessed speech processing in noise, and otoacoustic emissions assessed the physiological state of the cochlea.
Pure-tone thresholds were measured in 5 dB steps at six audiometric frequencies (0.25, 0.5, 1, 2, 4, and 8 kHz). Audiometry was conducted using custom software written in MATLAB (MathWorks) that implemented a method of adjustment psychophysical paradigm [
Distribution of pretest audiometric PTA for each group. The
Distribution of pretest audiometric HFA for each group. Group means: 45.8, 44.8, 38.6 dB HL for treatment, placebo, and control groups, respectively. Figure format is the same as described in Figure
The Hearing in Noise Test (HINT) [
Distribution of pretest HINT SNR50 scores for each group. Group means: 2.1, 0.7, and −0.5 dB for treatment, placebo, and control groups, respectively. These values were used to set the signal-to-noise ratio for both the CST pre- and posttests. Figure format is the same as described in Figure
The CST [
Transient-evoked otoacoustic emissions (TEOAEs) are physiologic measures of the cochlea’s response to a click-like acoustic stimulus [
TEOAEs were filtered into two bands: 1-2 kHz and 2–8 kHz. These frequency bands were chosen to correspond with the audiometric analyses and in this paper are referred to as the TEOAE PTA (1-2 kHz) and the TEOAE HFA (2–8 kHz). The distribution of pretest TEOAE amplitudes is shown for each frequency band in Figures
Distribution of pretest TEOAE PTA amplitudes (1-2 kHz).
Distribution of pretest TEOAE HFA (2–8 kHz) amplitudes. Figure format is the same as described in Figure
Data were obtained from both ears of each subject. Since no obvious differences were seen between left and right ears, data from both ears were combined in the following analyses. Strictly speaking, this likely violates the statistical assumption of independent sampling, since the test results from left and right ears of a single subject are likely to be highly correlated. None of the statistical tests used in the analyses are robust to the assumption of independent sampling, and the effect of including both ears is likely to be that of artificially increasing the sample size, making it more likely that a statistically significant result will be found. All statistical tests were conducted using a significance level of
Changes in the low-frequency pure-tone thresholds (PTA) were calculated by subtracting each subject’s pretest PTA from their posttest PTA. Changes in the high-frequency pure-tone thresholds were computed in the same way using the HFA thresholds. Negative values indicated an improvement in thresholds after treatment, and positive values indicated a worsening. Figures
Change in the audiometric PTA. Change calculated as posttest minus pretest; negative values indicate improvement in thresholds.
Change in the audiometric HFA. Change calculated as posttest minus pretest; negative values indicate improvement in thresholds. Figure format is the same as described in Figure
Changes in PTA and HFA across the three groups were compared statistically. Analysis of variance was used to test the null hypothesis that the population means of the groups are all equal. Use of ANOVA requires four assumptions: (1) data are from groups with normally distributed populations; (2) data are from groups with equal population variances; (3) groups are independent; (4) data within groups are independent and randomly sampled. The test is robust to the first and second assumptions if the number of samples in each group is large and equal or nearly equal. The test is never robust to the third and fourth assumptions. The sample sizes in this data set (
An analysis of variance showed no difference between group means for changes in PTA (
ANOVA for difference between audiometric PTA group means.
Source | SS | df | MS |
|
|
---|---|---|---|---|---|
Between | 2.03 | 2 | 1.02 | 0.091 | 0.913 |
Within | 632.95 | 57 | 11.10 | ||
| |||||
Total | 634.99 | 59 |
ANOVA for difference between audiometric HFA group means.
Source | SS | df | MS |
|
|
---|---|---|---|---|---|
Between | 24.78 | 2 | 12.39 | 1.326 | 0.274 |
Within | 532.40 | 57 | 9.34 | ||
| |||||
Total | 557.18 | 59 |
Before computing changes in CST performance, scores were first transformed into rationalized arcsine units (rau) [
Change in CST scores expressed as rationalized arcsine units. Change calculated as posttest minus pretest; positive values indicate improvement in speech intelligibility. Figure format is the same as described in Figure
Changes in rau scores across the three groups were compared statistically. Analysis of variance was used to test the null hypothesis that the population means of the groups are all equal. The assumptions required by ANOVA were discussed previously. As they apply to the rau difference data, the sample sizes were probably large enough and close enough to the same size to meet the first two assumptions. However, the sampled data do suggest the possibility that the groups are differently skewed (sk = 0.471, −1.40, −0.55 for the treatment, placebo, and control groups, resp.). A Kruskal-Wallis test was therefore also performed to compare the medians of the groups. The Kruskal-Wallis test technically requires the assumption that the populations of the different groups are identical. The test is robust to all differences except differences in variability between groups. The test is reasonably robust to differences in variability if the sample sizes are equal. While the sample sizes in this data set were not exactly equal, they were close to the same. Additionally, the standard deviations, which are reasonable estimates of variability, were reasonably similar (SD = 13.06, 16.04, 10.55 for the treatment, placebo, and control groups, resp.).
An analysis of variance showed no difference between group means for changes in rau score (
ANOVA for difference between CST group means, expressed in rationalized arcsine units.
Source | SS | df | MS |
|
|
---|---|---|---|---|---|
Between | 782.61 | 2 | 391.30 | 2.204 | 0.120 |
Within | 10121.90 | 57 | 177.57 | ||
| |||||
Total | 10904.52 | 59 |
Kruskal-Wallis test for difference between CST group medians, expressed in rationalized arcsine units. This test was performed on scores transformed to ranks. The ranks assigned to tied scores were the average of the ranks those scores would have had if they were not tied. Adjusted scores (adj.) are also shown. The adjustment is based on the fact that the variance of the ranks is smaller when ties are present. Its use is justifiable in cases where ties can be assumed to be present in the population. Here, it is assumed that many of the ties are due to rounding. The adjustment is small unless there are large numbers of tied scores.
Source | SS | df | Kw |
|
Kw (adj.) |
|
---|---|---|---|---|---|---|
Between | 123.08 | 2 | 4.036 | 0.133 | 4.063 | 0.131 |
Changes in the lower-frequency TEOAE amplitudes were calculated by subtracting each subject’s TEOAE PTA obtained during the pretest from their TEOAE PTA obtained during the posttest. Changes in the higher-frequency TEOAE amplitudes (TEOAE HFAs) were computed in the same way. Because TEOAEs are generated as a byproduct of outer hair cell function, significant positive values would theoretically be indicative of an improvement in outer hair cell function after treatment, and significant negative values would indicate a worsening. Figures
Change in TEOAE PTA amplitudes (1-2 kHz). Change calculated as posttest minus pretest; positive values indicate improvement. Figure format is the same as described in Figure
Change in TEOAE HFA (2–8 kHz) amplitudes. Change calculated as posttest minus pretest; positive values indicate improvement. Figure format is the same as described in Figure
Changes in TEOAE PTA and TEOAE HFA across the three groups were compared statistically. Analysis of variance was used to test the null hypothesis that the population means of the groups are all equal. Regarding the assumptions required by ANOVA, the smaller sample sizes of the TEOAE data set were probably not large enough to make the test robust to the assumption of normality. The groups in the TEOAE PTA data set (Figure
There is no theoretical reason to expect the higher-frequency TEOAE data to be distributed differently from the lower-frequency data; however, the groups in the TEOAE HFA data set (Figure
Analysis of variance showed no difference between group means for changes in TEOAE PTA (
ANOVA for difference between TEOAE PTA group means.
Source | SS | df | MS |
|
|
---|---|---|---|---|---|
Between | 2.18 | 2 | 1.09 | 0.133 | 0.876 |
Within | 253.96 | 31 | 8.19 | ||
| |||||
Total | 256.14 | 33 |
ANOVA for difference between TEOAE HFA group means.
Source | SS | df | MS |
|
|
---|---|---|---|---|---|
Between | 1.92 | 2 | 0.96 | 0.200 | 0.819 |
Within | 148.23 | 57 | 4.78 | ||
| |||||
Total | 150.15 | 59 |
Kruskal-Wallis test for difference between CST group medians, expressed as rationalized arcsine units. The ranks were assigned as described in Table
Source | SS | df | Kw |
|
Kw (adj.) |
|
---|---|---|---|---|---|---|
Between | 160.32 | 2 | 1.617 | 0.446 | 1.617 | 0.446 |
None of the three measures of hearing (audiometric thresholds, speech recognition test, or otoacoustic emissions) showed a statistically significant difference between the treatment, placebo, or control groups. Although a statistically significant difference between the groups might be detected with a larger sample size, such statistical significance may not necessarily be clinically meaningful. As discussed later, no individuals showed any clinically significant changes on any of the auditory tests.
From a clinical standpoint, a pure-tone threshold change of 10 dB or greater is generally considered significant [
When considering clinically significant changes in CST scores, it is necessary to know the critical difference of the scores, expressed in rau. Cox et al. [
From Figure
When considering clinically significant change in TEOAEs, an amplitude change of 6 dB or greater might be considered significant given the test-retest reliability in normal populations [
As described in Section
As discussed previously, several statistical assumptions must hold true in order to report valid statistics (normality, equal variance, independence, and random sampling). Since the current study was intended to be a feasibility study, it was anticipated that by randomly sampling individuals with documented sensorineural hearing loss, some evidence of an intervention effect would be measureable, if any existed. Since no effect could be demonstrated across a number of outcomes for any individual subjects, the study was terminated.
In this feasibility study, the timeline was fixed as per the pilot data from the manufacturer of the device. It is possible that the treatment number, treatment protocol, or even the duration of the entire data gathering was insufficient to show an intended effect.
The laser device was checked weekly, as per the manufacturer’s guidelines. It is possible, though unlikely, that the laser diode or the total output power varied between subject applications, all of which took place within a 7–10-day period.
No statistically significant effect of LLLT on auditory function was found, as assessed by pure-tone audiometry, speech understanding, and TEOAEs. Additionally, no individual subjects showed any clinically significant change. It remains possible that other methods of LLLT could have an effect on hearing. Further research elucidating the anatomic and physiologic bases for therapeutic effects of LLLT on hearing are needed before further clinical testing is warranted.