Introduction

Sudden sensorineural hearing loss (SSNHL) is an acute, unexplained hearing loss of at least 30 dB over at least three contiguous frequencies occurring within 72 h. A variety of treatments have been described for this condition, including vasoactive substances, hyperbaric oxygen, antivirals, vitamins, and even zinc [1–16]. However, ever since the 1980s when two double-blind trials [17, 18] showed efficacy of corticosteroids in the treatment of this condition, they have become the most commonly used agents in most centers worldwide, albeit with controversy. The systemic routes of administration (oral or intravenous) are often used.

There are patients who do not respond sufficiently to this mode of treatment. For this group of patients, some studies have demonstrated benefit in the use of salvage intratympanic steroids [19, 20], whereas others have demonstrated no additional benefit [21–24]. In view of the lack of agreement of multiple retrospective and prospective studies, some authors have conducted randomized controlled trials (RCTs) on this subject. However, these RCTs are limited by their inability to obtain a sufficiently large sample size. To date, there has been no meta-analysis done on RCTs to investigate the efficacy of salvage intratympanic steroids. The aim of this study was to pool and perform a meta-analysis on all relevant RCTs done on this topic, to (1) evaluate the efficacy of salvage intratympanic steroid injections in treating SSNHL; and (2) determine the type of steroid, dose of steroid, and method of administration that has been used with most success.

Methodology

Search strategy and selection criteria

A MEDLINE literature search was performed using a combination of the low-specificity keywords “hearing loss”, “steroid”, and “intratympanic”, supported by searches of Web of Science, Biosis, and Science Direct, to yield all possibly relevant results. The search was completed in May 2014. Articles of all languages were included.

We sought all RCTs that studied the efficacy of salvage intratympanic steroids in patients with SSNHL who have failed systemic steroid treatment. All RCTs fulfilling the following criteria were included: (1) conduct of human studies involving subjects with SSNHL who have failed systemic steroid treatment; (2) presence of a control arm (where no further treatment was prescribed) and a treatment arm (where salvage intratympanic steroids were given); (3) the average hearing threshold of each arm of the study was reported at the start and end of the treatment/observation period; (4) steroid treatment regimen was described.

Studies were excluded if they had incomplete reporting of pure-tone audiometry results pre- or post-intervention as this information was needed to calculate effect size. Attempts to obtain the required information from the authors were made, and these studies were only excluded if these attempts were unsuccessful.

All articles were de-identified (blinded title, authors, journal name, and year of publication) before selection. Selection of relevant publications was conducted independently by three authors, and any disagreements were resolved through discussions. The following information was extracted from each article: sample size of each study arm, mean age of the study group, gender distribution of the study group, type of steroid used, method of administering steroids, dose of steroid used, duration of therapy in treatment arm, and finally, pure-tone audiometry threshold pre- and post-study.

Statistical analysis

All statistical analyses were performed with Comprehensive Meta-Analysis Version 2.0, developed for support in meta-analysis. Meta-analysis of change scores using the random-effects model was performed. The random-effects model was used because it takes into account both variation caused by sampling error and also random variation of the underlying effect sizes between studies. A fixed-effect model would produce a confidence interval that may be artificially narrow as it only reflects the random variation within each trial, but not the potential heterogeneity between trials [25, 26]. Change in pure-tone audiometry scores between patients who did and did not receive salvage intratympanic steroids were calculated using a standardized mean difference (SMD) [27], together with its confidence interval and p value. Significant difference was set at p < 0.05 for all analyses. Tests of heterogeneity were conducted with the Q statistic that is distributed as a χ 2 variate under the assumption of homogeneity of effect sizes. Between-study heterogeneity was assessed with the I 2 statistic, which describes the percentage of variability among effect estimates beyond that expected by chance [28]. As a reference, I 2 values of 25 % were considered low, 50 % moderate, and 75 % high. Subgroup analyses were performed to investigate the role of steroid type and administrative method on outcome. Funnel plots and statistical tests (Egger’s linear regression method) [29] for funnel plot asymmetry were performed to test for evidence of publication bias.

Results

From the initial 184 articles found via the search strategy, 6 studies fulfilled the inclusion criteria. One study did not report relevant data to calculate effect size (Fig. 1). The five studies were published between the years 2006 and 2011. Two studies were conducted in Western countries and three were conducted in Asian countries. All studies involved 203 patients with SSNHL who have failed systemic steroid treatment. One hundred and two patients underwent further treatment with intratympanic steroids (i.e. cases) and 101 patients received no further therapy (i.e. controls). Other characteristics of the five studies are summarized in Table 1. Data from the five studies were pooled for meta-analysis. Results showed that patients who received salvage intratympanic steroids demonstrated a statistically significant reduction in the hearing thresholds (SDM = −0.401, SE = 0.143, 95 % CI −0.68 to −0.122, p = 0.005) as compared to controls, reflecting a greater amount of hearing improvement. Figure 2 shows the Forest plot and the standard mean difference in reduction of hearing thresholds in patients receiving salvage intratympanic steroids versus controls. No between-study heterogeneity was found (τ 2 = 0.000, Q = 2.751, df = 4, p = 0.600, I 2 = 0). As a result, meta-regression was not performed. We undertook subgroup analyses to explore the relationship between the mode of administration and type of steroid on the hearing thresholds as compared to controls (Table 2). The subgroup analysis showed that administration by injection (SDM = −0.375, p = 0.013) rather than a catheter (SDM = −0.629, p = 0.160) caused significant reduction in hearing thresholds or grater magnitude in improvement. The use of dexamethasone (SDM = −0.379, p = 0.039) rather than methylprednisolone (SDM = −0.459, p = 0.187) caused significant reduction in hearing thresholds.

Fig. 1
figure 1

Literature search profile

Full size image
Table 1 Study characteristics
Full size table
Fig. 2
figure 2

Standard difference in mean amount of change in hearing threshold (reduction equals improvement) in patients receiving further steroids versus controls

Full size image
Table 2 Subgroup analysis based on type of steroid and mode of administration
Full size table

Side effects of intratympanic steroids were reported by four of the five studies. Minor side effects included transient dizziness, ear pain, and tinnitus. Of the 203 patients in these studies, three developed tympanic membrane perforation. Of the three, one healed spontaneously, one was treated successfully with a paper patch, and one required a myringoplasty (in this patient a round window catheter was used). No infective complications occurred. The presence of publication bias was tested by the Egger’s regression method. There was no publication bias (intercept = −2.82, 95 % CI −9.71 to 4.07, t = 1.30, df = 3, p = 0.28).

Discussion

While there have been systemic reviews and meta-analyses previously done on the treatment of sudden sensorineural hearing loss, none has been performed specifically on RCTs evaluating the efficacy of salvage intratympanic steroids in patients who have previously failed systemic steroid treatment. A systemic review published by Spear et al. in 2011 [30] did include a section on this subject. However, a mixture of prospective studies and RCTs was used in their meta-analysis, which also did not include three randomized controlled trials [21, 22, 24] that were published in 2011, as their literature search was completed earlier. With the three additional RCTs published in 2011, it is now feasible to pool these studies with two previous RCTs [19, 23] to perform a new meta-analysis that only includes RCTs on this subject, so as to derive better evidence.

There have been previous meta-analyses done on related topics that we chose not to reexplore in this study. For example, Conlin et al. [32], Wei et al. [31], and Labus et al. [33] performed meta-analyses to evaluate the efficacy of systemic steroids versus no treatment in sudden sensorineural hearing loss. None of the three meta-analyses showed a statistically significant improvement in outcome when patients with sudden sensorineural hearing loss were treated with systemic steroids. However, the presence of individual RCTs [17, 18] that show the contrary, and the low incidence reported of adverse outcomes associated with treatment makes it a common practice for most centers in the world to treat these patients with systemic steroids nonetheless. Other studies have been performed to evaluate the efficacy of primary treatment with intratympanic steroids [34–38] or with combined systemic and intratympanic steroids [34, 39–44]. This meta-analysis did not include these studies, as it is more difficult to justify the first-line use of more invasive treatment before a trial of medical therapy.

This meta-analysis found that salvage intratympanic steroids is superior to no further treatment in patients with sudden sensorineural hearing loss who have failed systemic systemic steroids. Failure of systemic steroids was defined by the studies included as either (1) pure-tone average of worse than 30 dB; or (2) worse than 10–20 dB from the contralateral ear. Patients failing systemic steroid therapy by this criteria are therefore ideal candidates to be considered for intratympanic steroid therapy. Subgroup analysis showed that administration of dexamethasone via intratympanic injections yields the best outcomes. We did not analyze the effect of duration of salvage intratympanic steroid therapy on outcomes as this was 14–15 days for all five studies. Also, apart from the one study where dexamethasone was administered continuously via a round window catheter, all studies administered the intratympanic steroid injections four times over the treatment period, at a dose of 20 mg/injection (methylprednisolone) or 1.5–2 mg/injection (dexamethasone). In all studies, intratympanic steroids were performed within 1 month of the onset of sudden sensorineural hearing loss, after systemic steroids were completed. Patients who received intratympanic injections were all instructed post-injection to keep their heads still and turned to the opposite side for 20–45 min, and to refrain from swallowing in that time. The above measures may serve as a useful guide for clinicians performing intratympanic steroid injections.

This study has several strengths. First, there was no heterogeneity and publication bias in our results. Second, as a meta-analysis of RCTs, it provides a good level of evidence that salvage intratympanic steroids are efficacious in the treatment of refractory sudden sensorineural hearing loss. The main limitation of this meta-analysis is the small number of trials involved. This is related to the nature of the topic as only patients with refractory sudden sensorineural hearing loss are considered; the meta-analysis mentioned above performed by Spear et al. [30] also included a similar number of trials. Also, although all the studies included were randomized controlled trials, only one was blinded and placebo controlled. Although subgroup analysis found that dexamethasone rather than methylprednisolone, administered via intratympanic injections rather than a round window catheter tended to demonstrate better outcomes, we should view this result as preliminary. Although this study yielded statistical significance, the degree of clinical significance is still debatable due to the limitations mentioned above. Also, the studies included in this meta-analysis did not report the correlation statistic between pre and post-treatment audiometry scores. As such, we are unable to calculate the weighted raw mean difference in dB improvement in patients who underwent salvage intratympanic steroids. Should all five studies be assigned the same weight, patients who underwent salvage intratympanic steroids were found to have improved a mean of 10.0 dB more than patients who did not. The significance of this amount of improvement is debatable.

Salvage intratympanic steroids is currently not routinely practiced in many otology centers. Although it is acknowledged that the results of this study should be interpreted with appropriate caution, it is hoped that this study would encourage more clinicians to consider the use of this modality of treatment in patients who have failed initial systemic steroid treatment. With more widespread use of intratympanic steroids, more robust evidence may be generated of its efficacy.

Conclusion

In conclusion, this meta-analytical review provides evidence that for patients who have failed initial treatment with systemic steroids, salvage intratympanic steroid injections demonstrate statistically significant improvement and reduction in the hearing thresholds as compared to controls. The subgroup analysis showed that administration by injection rather than a catheter or the use of dexamethasone rather than methylprednisolone caused more significant reduction in hearing thresholds or grater magnitude in improvement. Clinicians may consider the use of salvage intratympanic dexamethasone injections in patients who have experienced treatment failure with systemic steroids.