Customize your JAMA Network experience by selecting one or more topics from the list below.
KAREN H.CALHOUNMDRONALD B.KUPPERSMITHMDFrom the Division of Otology/Neurotology, Department of Otolaryngology–Head and Neck Surgery, University of Michigan Health System, Ann Arbor.
Sudden-onset unilateral sensorineural hearing loss has no effective treatment.
The phenomenon of sudden-onset, unilateral sensorineural hearing loss (SSNHL) has been recognized and well described for some time. Whether it represents a single pathophysiologic entity or is the common end point of numerous, varied pathologic processes remains unknown. Many different causes of SSNHL have been hypothesized: viral infection of the labyrinth or cochlear nerve; vascular insult; perilymphatic hypoxia; intralabyrinthine membrane rupture; inflammatory and metabolic causes; and others. However, none has been shown to cause all, or even most, cases of SSNHL. There is not even universal acceptance of a standard definition of SSNHL. Although many authors define SSNHL as loss of at least 30 decibels (dB) in 3 contiguous frequencies over a period of 3 days or less, some investigators use more or less stringent criteria. This article reviews the most significant published data on treatment of SSNHL and attempts to synthesize them and draw reasonable conclusions on optimal treatment of this difficult problem.
David J. Eisenman, MD
There are many factors that render efficacy of treatment for SSNHL difficult to study. All of the most thorough studies of the natural history of SSNHL have documented a high rate of spontaneous resolution, at least in certain subsets of patients. This rate is usually estimated at about 65%. Because of the high rate of spontaneous resolution, well-controlled studies with comparisons of patients randomized to treatment and placebo groups are essential for documentation of treatment-related improvement.1,2 Many variables have been consistently shown to affect the rate of spontaneous resolution in patients with SSNHL. Severity of the loss (as measured by the pure-tone average), flat or downsloping audiograms, and the presence of vertigo are generally predictors of a poorer outcome. Other possible prognostic factors include age (very young and very old patients do worse), time from onset to initial diagnosis, and elevated erythrocyte sedimentation rate. In general, patients with lower frequency hearing loss recover better. All of these factors must be accounted for when trying to ascertain the benefit of any specific treatment.
H. Alexander Arts, MD
Even when prognostic variables and spontaneous resolution are controlled for, assessment of recovery is difficult since most patients do not have a premorbid audiogram for comparison. Different outcome variables have been used, including analysis of absolute improvement in decibels, analysis of percent improvement relative to the unaffected ear, and semiquantitative grading of improvement (complete, good, fair, poor, or none) based on composite ipsilateral and contralateral audiometric findings. A reliable study of the therapeutic benefit of any treatment must account for these difficulties in its analysis.
Our limited understanding of the etiology of SSNHL has as its corollary the lack of an effective treatment. Without understanding why SSNHL occurs, it has been difficult if not impossible to develop an appropriate treatment. One approach has been to empirically administer to all patients a battery of treatments directed at all of the hypothesized causes. Such "shotgun" regimens typically include some or all of the following treatments: corticosteroids, vasodilators, diuretics, histamine, plasma expanders, carbogen inhalation (5% carbon dioxide + 95% oxygen), intravenous contrast dye, and calcium channel blockers. With such a broad, empiric regimen it is hoped that many of the potential causes of SSNHL are treated. Wilkins et al3 retrospectively reviewed the outcomes in 132 patients treated with a shotgun regimen. They compared the results in patients who received the entire treatment protocol with those who received less than the entire protocol and with historical controls. They found no difference in hearing results between patients receiving less than the entire protocol vs those receiving the entire protocol. Furthermore, there were no significant differences in outcome when absence of any single drug was analyzed. Also, the overall rate of hearing recovery was well within the limits of what could be attributed to spontaneous resolution, as would be expected from the natural history of the disease. Although not a prospective, controlled study, the results of this article suggest that none of the therapies for SSNHL employed in the study was effective.
Other authors have examined more specific therapies for SSNHL. Three studies have specifically examined the effectiveness of corticosteroids in the treatment of SSNHL and found no benefit, though none was performed in a prospective manner.4-6 In each of these 3 studies, all of the patients received other treatments in addition to corticosteroids. Furthermore, none of these studies included an untreated control group. In 2 of the studies, audiometric outcomes of patients receiving corticosteroids were compared with those of patients not receiving corticosteroids.5,6 Both groups demonstrated equivalent outcomes. One study was simply a retrospective review of patients treated with 3 agents, 1 of which was a corticosteroid.4 In this latter study, the overall improvement rate for the group of patients treated with the 3 agents (including a steroid) was compared with the expected rate of spontaneous resolution and found to be no better.
Many other treatment options besides corticosteroids have been proposed and anecdotally reported to be of benefit in SSNHL, but none has demonstrated effectiveness in well-designed studies.7-9 Most of the studies are deficient in failing to have adequate control groups (either retrospective or prospective) and/or sufficient numbers of patients to provide adequate statistical power. One example of this is a study that purports to demonstrate the effectiveness of "defibrinogenation."9 The authors randomized 169 patients to treatment with either intravenous corticosteroids or baxtrobin (a thrombinlike enzyme that decreases serum fibrinogen levels). There was no untreated control group. The overall improvement rate in the steroid group was 38.7% and in the baxtrobin group 57.3%. This difference was statistically significant when tested with a χ2 analysis.
When looking at these numbers objectively, however, the rate of improvement in the steroid group is worse than what is usually seen with spontaneous resolution alone, while the rate of improvement in the baxtrobin group is no better than what one would expect with no treatment. This casts doubt on the study's validity, and raises the possibility that it is not the baxtrobin group that is doing better than expected, but rather the steroid group that is doing worse for some unexplained reason. Without an untreated control group this problem cannot be resolved.
An analogous problem is found in Fisch's 1983 report10 on the efficacy of carbogen inhalation for SSNHL. A group of patients receiving carbogen inhalation treatment was compared with a group of patients receiving intravenous infusions of papaverine and dextran. Overall, patients receiving carbogen inhalation showed greater improvements in hearing, as compared with patients receiving papaverine and dextran infusions, when tested 1 year following treatment. However, absent an untreated control group, it is impossible to say whether it is the carbogen-treated patients who did better or the papaverine/dextran– treated patients who did worse. In fact, in the very same article, the author presents data demonstrating that papaverine infusion decreases perilymphatic oxygen tension, precisely the opposite effect of that which gives carbogen its hypothesized efficacy. The same problem is found in an article by Rahko and Kotti,11 the findings of which demonstrated no difference between patients treated with heparin infusion or carbogen inhalation. It is not known whether either group did any better (or worse) than would be expected with no treatment at all.
There are 2 prospective, randomized, double-blind, placebo-controlled trials that studied the efficacy of vasoactive drugs on audiometric outcomes in SSNHL. One study compared patients treated with a combination of intravenous dextran and procaine with an untreated placebo group12; the other study compared patients treated with either intravenous dextran and pentoxifylline or pentoxifylline alone with an untreated placebo group.13 Though the number of subjects in the former investigation by Kronenberg and coworkers12 limits the statistical power of their study, both studies failed to demonstrate improved outcomes in the treatment arms. Excepting some further data on corticosteroids that will presently be discussed, the myriad of purported treatments and persisting absence of improved outcomes suggest that there is no identifiable effective therapy for SSNHL.
The dearth of well-controlled studies of SSNHL gives unique weight to those that do exist. Of the 3 prospective, randomized, placebo-controlled, double-blind studies of SSNHL published to date, only 1 has examined the efficacy of the most commonly used drug, ie, corticosteroids.14 The other 2 such designed reports (reviewed above), which examined the efficacy of vasoactive and rheologic agents, found no benefit to the therapies that they studied.12,13 In 1980, Wilson et al14 reported the outcomes of 2 studies conducted concurrently at the Massachusetts Eye and Ear Infirmary in Boston, Mass, and the Kaiser-Permanente Medical Group in Oakland, Calif. Sudden sensorineural hearing loss was defined as hearing loss of at least 30 dB occurring in at least 3 contiguous frequencies in less than 3 days. In each study, eligible patients were randomized to receive either placebo (with no other treatment) or bolus and tapering doses of oral corticosteroids. Although the corticosteroids administered in the 2 studies differed (dexamethasone in California and methylprednisolone in Massachusetts), they had equipotent anti-inflammatory effects. Recovery was defined as complete if the posttreatment pure-tone average or speech reception threshold was within 10 dB of the premorbid level. Partial recovery was defined as recovery of 50% of the hearing loss. Patients were considered to have no recovery if they had less than 50% recovery of hearing in the affected ear. The premorbid hearing was considered equivalent to the hearing in the unaffected ear when no premorbid audiogram was available. Patients with partial and complete recovery were grouped together for the statistical analysis. Thirty-three patients received steroids and 34 received placebo. Fifty-two patients from one of the institutions refused randomization and were not treated at all (neither steroid nor placebo).
After controlling for other variables that affect outcome, such as age and presence of vertigo, the authors of the 2 reviewed studies found that patients treated with oral corticosteroids had statistically significant relative odds of 4.95:1 favoring recovery.14 The efficacy of steroids was limited to 1 of 3 groups of patients, based on the presenting audiogram configuration. All patients whose hearing loss was greatest in the midfrequencies, even when severe, and even when accompanied by vertigo, had complete recovery regardless of treatment. Of the patients who had hearing loss greater than 90 dB at all frequencies, 76% showed no recovery, and steroid treatment had no effect on their outcome. Patients in this group had a higher incidence of vertigo than did those in the other groups. Patients whose audiometric configurations placed them between these 2 extreme categories were those in whom steroids were found to be beneficial.
This study was very well designed, and most confounding variables were well controlled and accounted for. The relatively small size of the study and the relatively strict definition of recovery leave open the possibility of some falsely negative results, ie, the presence of an actual therapeutic effect where none was discernible. However, the positive results demonstrated are quite convincing. At least 2 retrospective studies of SSNHL since then have also suggested a beneficial effect of oral corticosteroid treatment, though each suffers from many of the flaws inherent in retrospective study of this disease.15,16
In the large, retrospective study of SSNHL by Fetterman et al,15 the authors examined not only the efficacy of steroids, but also of vasodilators. This latter category was a heterogeneous group of drugs including histamine, niacin, papaverine, and cyclandelate. The authors examined the outcomes of 242 patients according to whether they received steroids without vasodilators, vasodilators without steroids, both, or neither. Many patients also received other drug treatments. Recovery was judged by a 10-dB improvement in pure-tone average or 15% improvement in speech discrimination score. This method of outcome assessment, without consideration of how close to baseline the hearing recovered, may account for the surprising finding that patients with poorer initial hearing showed a greater rate of improvement, since they had more room to improve. The data seem to indicate that both steroids and vasodilators had independent effects on improving hearing outcomes. However, results in all of the groups, except the small group of patients who received neither steroids nor vasodilators, are well within the range of what could be expected from spontaneous recovery alone, especially considering the definition of improvement. Furthermore, the treatment groups are not broken down by other factors known to affect prognosis, so it is impossible to tell if the groups are comparable. In particular, the very poor outcome (7.7% improvement—far worse than would be expected from spontaneous recovery alone) in the group of patients who received neither steroids nor vasodilators suggests that this group is not a valid control for comparison with the treatment arms.
If we consider only the most reliable studies of SSNHL, the data suggest 2 important points: (1) A substantial number of patients with SSNHL will have spontaneous recovery of some or all of their hearing, depending on the presence of certain established prognostic indicators, as discussed above. (2) Oral corticosteroids are an effective treatment for patients with intermediate audiograms (ie, neither limited to midfrequency loss nor with profound loss at all frequencies). Therefore, the hypothesis that there is no effective treatment for patients with SSNHL should be rejected. This literature review suggests that the hypothesis should be reformulated to assert that steroids are an effective treatment for certain patients with SSNHL.
Many issues remain unsolved, however. Are steroids the only effective treatment? Are they the most effective treatment? What, if any, is the cause of SSNHL in most cases?
Two important treatment options were not discussed above, since they have not been the focus of any significant clinical trials. Patients ultimately diagnosed with Meniére disease are excluded from most studies of SSNHL, but it is well accepted that SSNHL (with or without vertigo) can represent a first attack of endolymphatic hydrops. Therefore, consideration should be given to administration of oral diuretics and recommendation of a low-salt diet to patients with SSNHL.
Evidence continues to accumulate suggesting that viral infection may play an important role in SSNHL. With the availability of relatively inexpensive, well-tolerated antiviral agents, many clinicians prescribe these drugs as treatment for SSNHL, either in combination with steroids or alone. The efficacy of antiviral agents is unproven, but it is currently being tested in a multicenter, prospective, randomized, controlled clinical trial. The results of that study may also help reveal the etiology of SSNHL.
Finally, although the incidence of acoustic neuroma in patients with SSNHL is low (1%-2% of all patients, at most), up to 10% of patients with acoustic neuroma present with SSNHL. For this reason, after initiation of corticosteroid treatment with or without an antiviral agent, strong consideration should be given to magnetic resonance imaging, or possibly auditory brainstem response testing, to exclude a retrocochlear lesion as the cause of SSNHL. Improvement in hearing with corticosteroid treatment does not exclude the possibility of a retrocochlear lesion, so response to treatment should not influence the decision to obtain magnetic resonance images or auditory brainstem response results. Although corticosteroids have demonstrated efficacy only for the group of patients described earlier, many clinicians treat all patients with SSNHL, unless individual contraindications exist, with the hope that this relatively well-tolerated treatment will be of some benefit.
Accepted for publication May 18, 2000.
Reprints are not available.
Corresponding author: H. Alexander Arts, MD, Department of Otolaryngology–Head and Neck Surgery, University of Michigan Hospital, 1500 E Medical Center Dr, 1904 Taubman, Ann Arbor, MI 48109-0312 (e-mail: firstname.lastname@example.org).
Eisenman D, Arts HA. Effectiveness of Treatment for Sudden Sensorineural Hearing Loss. Arch Otolaryngol Head Neck Surg. 2000;126(9):1161–1164. doi:10.1001/archotol.126.9.1161
Browse and subscribe to JAMA Network podcasts!
Create a personal account or sign in to: