Corticosteroides intratimpánicos para la hipoacusia neurosensorial súbita
Resumen
Antecedentes
La hipoacusia neurosensorial súbita idiopática (HNSSI) es frecuente y se define como una disminución repentina de la sensibilidad auditiva neurosensorial de etiología desconocida. Los corticosteroides sistémicos se utilizan ampliamente, pero su utilidad todavía no está clara. Las inyecciones intratimpánicas de corticosteroides se han vuelto cada vez más habituales en el tratamiento de la HNSSI.
Objetivos
Evaluar los efectos de los corticosteroides intratimpánicos en personas con HNSSI.
Métodos de búsqueda
El documentalista del Grupo Cochrane de Enfermedades de oído, nariz y garganta (Cochrane ENT) realizó búsquedas en el registro de ensayos de este Grupo; en CENTRAL (2021, número 9); PubMed; Ovid Embase; CINAHL; Web of Science; ClinicalTrials.gov; ICTRP y otras fuentes adicionales de ensayos publicados y no publicados (fecha de búsqueda: 23 de septiembre de 2021).
Criterios de selección
Se incluyeron los ensayos controlados aleatorizados (ECA) con personas con HNSSI y un seguimiento de más de una semana. Los corticosteroides intratimpánicos se administraron como tratamiento primario o secundario (tras el fracaso del tratamiento sistémico).
Obtención y análisis de los datos
Se utilizaron los métodos estándar de Cochrane, incluido el uso del método GRADE para evaluar la certeza de la evidencia. El desenlace principal de esta revisión fue el cambio del umbral de audición con la audiometría de tonos puros. Los desenlaces secundarios incluyeron la proporción de personas cuya audición mejoró, el umbral auditivo final, la audiometría del habla, los cambios auditivos específicos de la frecuencia y los efectos adversos.
Resultados principales
Se incluyeron 30 estudios que analizaron a 2133 participantes. Algunos estudios tenían más de dos grupos de tratamiento y, por tanto, eran relevantes para varias comparaciones. Los estudios investigaron los corticosteroides intratimpánicos como tratamiento primario (inicial) o secundario (de rescate) tras el fracaso del tratamiento inicial.
1. Corticosteroides intratimpánicos versus corticosteroides sistémicos como tratamiento primario
Se identificaron 16 estudios (1108 participantes). El tratamiento intratimpánico podría dar lugar a una mejoría escasa o nula en el cambio del umbral de audición (diferencia de medias [DM] ‐5,93 dB mejor; intervalo de confianza [IC] del 95%: ‐7,61 a ‐4,26; diez estudios; 701 participantes; certeza baja). Se encontró poca o ninguna diferencia en la proporción de participantes cuya audición mejoró (razón de riesgos [RR] 1,04; IC del 95%: 0,97 a 1,12; 14 estudios; 972 participantes; certeza moderada). El tratamiento intratimpánico podría dar lugar a una diferencia escasa o nula en el umbral auditivo final (DM ‐3,31 dB; IC del 95%: ‐6,16 a ‐0,47; siete estudios; 516 participantes; certeza baja). El tratamiento intratimpánico podría aumentar el número de personas que presentan vértigo o mareo (RR 2,53; IC del 95%: 1,41 a 4,54; un estudio; 250 participantes; certeza baja) y probablemente aumenta el número de personas con dolor de oído (RR 15,68; IC del 95%: 6,22 a 39,49; dos estudios; 289 participantes; certeza moderada). También provocó perforación persistente de la membrana timpánica (rango: 0% a 3,9%; tres estudios; 359 participantes; certeza muy baja), vértigo/mareo en el momento de la inyección (1% a 21%, tres estudios; 197 participantes; certeza muy baja) y dolor de oído en el momento de la inyección (10,5% a 27,1%; dos estudios; 289 participantes; certeza baja).
2. Corticosteroides intratimpánicos más sistémicos (tratamiento combinado) versus corticosteroides sistémicos solos como tratamiento primario
Se identificaron diez estudios (788 participantes). El tratamiento combinado podría tener un pequeño efecto en el cambio del umbral de audición (DM ‐8,55 dB mejor; IC del 95%: ‐12,48 a ‐4,61; seis estudios; 435 participantes; certeza baja). Es muy incierta la evidencia en cuanto a si el tratamiento combinado cambia la proporción de participantes con mejoría de la audición (RR 1,27; IC del 95%: 1,15 a 1,41; diez estudios; 788 participantes; certeza muy baja). El tratamiento combinado podría dar lugar a umbrales auditivos finales ligeramente inferiores (más favorables), pero la evidencia es muy incierta y no está claro si el cambio sería importante para los pacientes (DM ‐9,11 dB; IC del 95%: ‐16,56 a ‐1,67; tres estudios; 194 participantes; certeza muy baja). Algunos efectos adversos sólo se produjeron en los que recibieron el tratamiento combinado. Estos incluyeron la perforación persistente de la membrana timpánica (rango: 0% a 5,5%; cinco estudios; 474 participantes; certeza muy baja), vértigo o mareo en el momento de la inyección (rango: 0% a 8,1%; cuatro estudios; 341 participantes; certeza muy baja) y dolor de oído en el momento de la inyección (13,5%; un estudio; 73 participantes; certeza muy baja).
3. Corticosteroides intratimpánicos versus ningún tratamiento o placebo como tratamiento secundario
Se identificaron siete estudios (279 participantes). El tratamiento intratimpánico podría tener un pequeño efecto en el cambio del umbral de audición (DM ‐9,07 dB mejor; IC del 95%: ‐11,47 a ‐6,66; siete estudios; 280 participantes; certeza baja). El tratamiento intratimpánico podría dar lugar a una proporción mucho mayor de participantes con mejoría de la audición (RR 5,55; IC del 95%: 2,89 a 10,68; seis estudios; 232 participantes; certeza baja). El tratamiento intratimpánico podría dar lugar a umbrales auditivos finales más bajos (más favorables) (DM ‐11,09 dB; IC del 95%: ‐17,46 a ‐4,72; cinco estudios; 203 participantes; certeza baja). Algunos efectos adversos sólo se produjeron en los que recibieron la inyección intratimpánica. Estos incluyeron la perforación persistente de la membrana timpánica (rango: 0% a 4,2%; cinco estudios; 185 participantes; certeza muy baja), vértigo o mareo en el momento de la inyección (rango: 6,7% a 33%; tres estudios; 128 participantes; certeza muy baja) y dolor de oído en el momento de la inyección (0%; un estudio; 44 participantes; certeza muy baja).
4. Corticosteroides intratimpánicos más sistémicos (tratamiento combinado) versus corticosteroides sistémicos solos como tratamiento secundario
Se identificó un estudio con 76 participantes. No se informó acerca del cambio en el umbral de audición. El tratamiento combinado podría dar lugar a una mayor proporción de mejoría de la audición, pero la evidencia es muy incierta (RR 2,24; IC del 95%: 1,10 a 4,55; certeza muy baja). Los efectos adversos se informaron de manera deficiente, con datos sólo de la perforación persistente de la membrana timpánica (tasa 8,1%, certeza muy baja).
Conclusiones de los autores
La mayoría de la evidencia de esta revisión es de certeza baja o muy baja, por lo que es probable que estudios adicionales pudieran cambiar las conclusiones.
Para el tratamiento primario, los corticosteroides intratimpánicos podrían tener poco o ningún efecto en comparación con los corticosteroides sistémicos. Podría haber un ligero efecto beneficioso del tratamiento combinado en comparación con el tratamiento sistémico solo, pero la evidencia es incierta.
En cuanto al tratamiento secundario, hay evidencia de certeza baja de que los corticosteroides intratimpánicos, en comparación con ningún tratamiento o placebo, podrían dar lugar a una proporción mucho mayor de participantes con mejoría de la audición, pero podrían tener sólo un pequeño efecto en el cambio del umbral de audición. No está claro si existe un efecto beneficioso adicional del tratamiento combinado sobre los corticoides sistémicos solos.
Aunque los efectos adversos se informaron de manera deficiente, a la hora de seleccionar el tratamiento adecuado se deben tener en cuenta los diferentes perfiles de riesgo del tratamiento intratimpánico (que incluyen la perforación de la membrana timpánica, el dolor y los mareos/vértigo) y del tratamiento sistémico (por ejemplo, problemas de glucemia).
PICO
Resumen en términos sencillos
Tratamiento de la pérdida auditiva súbita con corticoides aplicados en el oído medio
¿Qué es la pérdida auditiva súbita?
La pérdida auditiva súbita es un trastorno que se caracteriza por la reducción repentina (normalmente en el transcurso de 72 horas) de la audición o la pérdida de la misma.
¿Cómo se trata?
A menudo se han utilizado corticoides (un tipo de medicamento antiinflamatorio) para tratar la enfermedad. Estos medicamentos se suelen tomar por vía oral o inyectarse en el organismo (conocidos como corticoides sistémicos), pero también se pueden administrar en forma de inyección directamente en el oído medio, a través del tímpano (conocidos como corticoides intratimpánicos).
¿Qué se quería averiguar?
No está claro si el tratamiento intratimpánico con corticoides es eficaz, ni cuál de estos tratamientos (intratimpánico o sistémico) es el mejor para tratar esta afección.
¿Qué se hizo?
Se buscaron todos los estudios relevantes en la literatura médica, se compararon sus resultados y se resumió la evidencia. La certeza de la evidencia también se evaluó considerando factores como el tamaño de los estudios y la forma en que se realizaron. Según las evaluaciones, la evidencia se calificó como de certeza muy baja, baja, moderada o alta.
¿Qué se encontró?
Se encontraron 30 estudios que incluyeron a 2133 personas. Estos estudios compararon el tratamiento intratimpánico con corticoides con ningún tratamiento, con placebo (tratamiento simulado o ficticio) y con corticoides que se tomaban por vía oral o se inyectaban en el cuerpo (corticoides sistémicos). Se tuvo en cuenta si las personas recibían un primer tratamiento para la sordera súbita o si habían recibido previamente algún otro tipo de tratamiento (que no había funcionado).
Para las personas que reciben un primer tratamiento de sordera súbita
No se encontraron estudios que compararan los corticoides intratimpánicos con ningún tratamiento o con un tratamiento placebo (ficticio).
Los corticoides intratimpánicos podrían dar lugar a una diferencia mínima o nula en la audición en comparación con las personas que reciben corticoides sistémicos, y podrían dar lugar a una diferencia mínima o nula en el número de personas con mejoría de la audición. Los efectos secundarios podrían ser diferentes con estos dos tipos de tratamiento. Con el tratamiento intratimpánico, las personas podrían tener un mayor riesgo de mareos o dolor de oído en comparación con los corticoides sistémicos, normalmente en el momento de la inyección, y algunas podrían desarrollar un pequeño agujero en el tímpano. Sin embargo, el tratamiento sistémico también podría causar un mayor riesgo de diferentes efectos secundarios, como problemas con los niveles de glucosa en la sangre.
Tomar corticoides intratimpánicos además de corticoides sistémicos podría dar lugar a una pequeña mejoría de la audición en comparación con los corticoides sistémicos solos, pero no se sabe con certeza cuántas personas notarían una mejoría. Al igual que en el caso anterior, el tratamiento intratimpánico podría causar algunos efectos secundarios, pero no es posible saber con certeza el número de personas que podrían experimentarlos.
Para las personas que reciben un tratamiento adicional para la sordera súbita (cuando el primer tratamiento no funcionó)
Cuando se comparan con ningún tratamiento o con un tratamiento placebo, los corticoides intratimpánicos podrían dar lugar a una mejoría de la audición en un número mucho mayor de personas, pero sólo mejorar ligeramente la audición. Al igual que con el primer tratamiento, las inyecciones intratimpánicas podrían provocar algunos efectos secundarios, como dolor o mareos en el momento de la inyección, o la aparición de un pequeño agujero en el tímpano. No está clara la frecuencia con la que se producen estos efectos secundarios.
No está claro que al agregar el tratamiento intratimpánico al tratamiento sistémico se vaya a producir una mejoría de la audición.
¿Cuáles son las limitaciones de la evidencia?
Se considera que la mayoría de la evidencia que se encontró es de certeza baja o muy baja. Esto se debe a que a menudo hubo algunos problemas con la forma en que se realizaron los estudios, se podrían haber incluido pocas personas en los estudios y a veces los resultados de diferentes estudios fueron contradictorios. Por lo tanto, las conclusiones de esta revisión podrían cambiar a medida que se publican nuevos estudios.
¿Cuál es el grado de actualización de esta evidencia?
La evidencia de esta revisión Cochrane está actualizada hasta el 23 de septiembre de 2021.
Authors' conclusions
Summary of findings
| Intratympanic corticosteroids versus systemic corticosteroids as primary therapy | ||||||
| Patient or population: sudden sensorineural hearing loss Settings: initial therapy Intervention: intratympanic steroid therapy Comparison: systemic steroid therapy | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | No of participants | Relative effect | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Systemic therapy (assumed risk) | Intratympanic therapy (corresponding risk) | |||||
| Change in hearing threshold determined by PTA
Range 0 dB to 140 dB
Negative values represent lowering and positive values represent raising of the hearing threshold. A lower hearing threshold represents hearing improvement).
| The mean change in PTA ranged across control groups from ‐30.07 dB to ‐15.1 dB | The mean change in PTA in the intervention groups was on average ‐5.93 dB greater (from ‐4.26 greater to ‐7.61 greater) | 701 (10 studies) | MD ‐5.93 dB (95% CI ‐7.61 to ‐4.26) | ⊕⊕⊝⊝ | Intratympanic therapy may have a trivial/no effect on the change in hearing threshold when compared to systemic steroids (as primary therapy). |
| Proportion of patients whose hearing is improved | 731 per 1000a | 760 per 1000 (709 to 818) | 972 (14 studies) | RR 1.04 (95% CI 0.97 to 1.12) | ⊕⊕⊕⊝ | Intratympanic therapy probably results in little to no difference in the proportion of patients whose hearing is improved compared to systemic corticosteroids (as primary therapy). |
| Final hearing threshold determined by PTA (a lower value represents better hearing) | The mean final PTA ranged across control groups from 25.1 dB to 59 dB | The mean final PTA in the intervention groups was on average ‐3.31 dB lower (‐6.16 lower to ‐0.47 lower) | 516 (7 studies) | MD ‐3.31 dB (95% CI ‐6.16 to ‐0.47) | ⊕⊕⊝⊝ low3 | Intratympanic therapy may result in little to no difference in the final hearing threshold (as primary therapy). |
| Adverse eventsb | Events in control group | Events in intervention group | No of participants (studies) | Relative effect (95% CI) | Certainty of the evidence (GRADE) | Comments |
| Tympanic membrane perforation | Comparison not applicablec | Ranged from 0% (0/30) to 3.9% (5/129) | 463 (4 studies) | Not calculable | ⊕⊝⊝⊝ very low4 | The evidence is very uncertain regarding the risk of tympanic membrane perforation for those who received intratympanic corticosteroid as primary treatment. |
| Vertigo/dizziness: timing not reportedd | 13/121 (10.7%) | 35/129 (27.1%) | 250 (1 study) | RR 2.53 (1.41 to 4.54) | ⊕⊕⊝⊝ low5 | Intratympanic therapy may increase the risk of vertigo/dizziness of unspecified timing as compared to systemic corticosteroid. |
| Vertigo/dizziness: at the time of injection | Comparison not applicablec | 3 studies reported a rate between 1.5% (1/67) and 21% (4/19) for those who received an intratympanic injectione | 301 (4 studies) | Not calculable | ⊕⊝⊝⊝ very low6 | The evidence is very uncertain regarding the risk of vertigo/dizziness at the time of intratympanic injection of corticosteroid as primary treatment. |
| Ear pain: timing not reportedf | 4/141 (2.8%) | 74/148 (50%) | 289 (2 studies) | RR 15.68 (95% CI 6.22 to 39.49) | ⊕⊕⊕⊝ | Intratympanic corticosteroid injection probably increases the risk of ear pain of unspecified timing as compared to systemic corticosteroid when used as primary treatment. |
| Ear pain: at the time of injectionf | Comparison not applicablec | 3 studies reported a rate between 4.8% (5/104) and 27.1% (35/129) | 393 (3 studies) | Not calculable | ⊕⊕⊝⊝ | The evidence suggests that there may be a risk of ear pain at the time of intratympanic injection of corticosteroid as primary treatment. |
| *The basis for the assumed risk is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| aFourteen studies recruited participants suffering from sudden sensorineural hearing loss. The incidence of improvement for the systemic corticosteroid group in these 14 studies was 73.07%. We have used 731 per 1000 to express the assumed risk. bOnly the most widely reported adverse events are described here. For adverse events that could feasibly occur in either group, we have only included the studies that provided a rate for both groups. For adverse events that could only occur in one group, we have only included the studies that reported the rate in that group, and presented these as a range. A full description of adverse event data is available for reference in Table 1. cComparisons between patients receiving intratympanic therapy and those receiving only systemic therapy were regarded as invalid for the following adverse events: persistent tympanic membrane perforation, vertigo observed at the time of intratympanic injection and ear pain observed at the time of intratympanic injection. This is explained in Data extraction and management. dA single study reported a rate for both intratympanic and systemic corticosteroid (Rauch 2011). However, it is not specified whether all of the patients in the intratympanic corticosteroid group experiencing vertigo did so at the time of injection. We have therefore reported this outcome separately from vertigo/dizziness interpreted as having occurred specifically at the time of injection. eIn two studies, two groups received intratympanic injection: in Tsounis 2018, one group received intratympanic corticosteroid and the other received intratympanic and systemic corticosteroid; in Huang 2021, one group received intratympanic corticosteroid and the other received intravenous followed by intratympanic corticosteroid. fIn each study contributing data, the number of participants with ear pain/earache was presented separately from the numbers with ear pain at intratympanic injection. It was assumed, therefore that those participants with pain at injection were not included among those with ear pain/earache.
GRADE Working Group grades of evidence | ||||||
| 1Downgraded one level due to risk of bias: eight studies were at high risk of other bias, three studies were at risk of attrition bias and three studies were at risk of selection bias. Downgraded one level due to inconsistency: the size and direction of effect varied between the studies and the I2 value was 80%. 2Downgraded one level due to risk of bias: we judged 11 of 14 studies to be at unclear or high risk of selection bias and we judged 12 of 14 studies to be at high risk of other bias. 3Downgraded one level due to risk of bias: we judged six studies to be at high risk of other bias; two studies were at high risk of selection bias. Downgraded one level due to inconsistency: the I2 value was moderate (41%). 4Downgraded one level due to risk of bias: we judged one study to be at high risk of bias because of concern about random sequence generation and allocation concealment. Downgraded two levels due to imprecision: the number of events is smaller than the optimal information size (taken as 300 events) and an effect estimate could not be calculated. 5Downgraded one level due to imprecision: the number of events is smaller than the optimal information size (taken as 300 events). Downgraded one level because of indirectness: provision of data by only a single study from a single setting, which may not adequately represent all patients with ISSNHL. 6Downgraded two levels due to risk of bias: we judged two studies to be at high risk of bias because of incomplete outcome data; we judged one study to be at high risk of bias because of concern about random sequence generation and allocation concealment. Downgraded two levels due to imprecision: the number of events is smaller than the optimal information size (taken as 300 events) and an effect estimate could not be calculated. 7Downgraded one level due to imprecision: the number of events is smaller than the optimal information size (taken as 300 events). 8Downgraded two levels because of imprecision: the number of events is smaller than the optimal information size (taken as 300 events) and an effect estimate could not be calculated. | ||||||
| Adverse event reported | Study | How reported | Details of recovery | Rate in intervention group (%) | Rate in comparator group (%) | RR (95% CI) |
|---|---|---|---|---|---|---|
| Tympanic membrane (TM) perforation | There were no cases of [...] perforation of the tympanic membrane | NA | 0/52 (0) | 0/52 (0)* | NA | |
|
| No residual TM perforations | All patients demonstrated a complete healing of TM after the tympanostomy tube removal | 0/24 (0) | NA | NA | |
|
| Persistent TM perforation | By the 6‐month follow‐up most adverse events had resolved | 5/129 (3.9) | NA | NA | |
|
| No residual tympanic membrane perforations were observed in any of the individuals at their final visit | NA | 0/30 (0) | NA | NA | |
| Vertigo/dizziness | Four patients in the ITS group had transient vertigo during the procedure | NR | 4/19 (21) | NA | NA | |
|
| [...] complained of brief dizziness after IT injection [...]
| No patients stopped the treatment | 7/52 (13.5) | 8/52 (15.4)* | NA | |
|
| The intratympanic group experienced adverse effects typical of local injection, most often transient pain at the injection site and brief caloric vertigo. Note, it is unclear whether all reported instances of vertigo in the intervention group occurred at the time of injection. | By the 6‐month follow‐up most adverse events had resolved | 35/129 (27.1) | 13/121 (10.7) | 2.53 (1.41 to 4.54); favours systemic corticosteroid; P = 0.002 | |
|
| Temporary adverse events in 22.7% of patients treated with oral prednisolone which included [...] and dizziness. In 35% of patients treated with intratympanic corticosteroid, adverse events occurred including [...] and dizziness. | NR | NR | NR | NA | |
|
| Six of 30 patients in the intratympanic injection group complained of a transient dizziness lasting about a minute during treatment | NR | 6/30 (20) | NA | NA | |
|
| One patient experienced transient dizziness as a result of caloric stimulation from the injected steroid solution (unclear which of 2 groups receiving intratympanic injection) | Symptoms resolved completely within 15 minutes and there was no need to discontinue the treatment. The injections that followed caused no further side effect. | NR | NR | NA | |
| Tinnitus | In 35% of patients treated with intratympanic corticosteroid, adverse events occurred including [...] ringing sensation in the ear | NR | NR | NR | NA | |
| Ear pain | Pain due to injection | NR | 2/19 (10.5) | NA | NA | |
|
|
| Earache | NR | 4/19 (21.1) | 0/20 (0) | 9.45 (0.54 to 164.49); favours systemic corticosteroid; P = 0.12 |
|
| [...] refused repeated IT injections due to unbearable pain | NR | 3/52 (5.8) | 2/52 (3.8) | NA | |
|
| The intratympanic group experienced adverse effects typical of local injection, most often transient pain at the injection site [...] | By the 6‐month follow‐up most adverse events had resolved | 35/129 (27.1) | NA | NA | |
|
|
| Experienced ear pain at least once | By the 6‐month follow‐up most adverse events had resolved | 70/129 (54.3) | 4/121 (3.3) | 16.41 (6.18 to 43.59); favours systemic corticosteroid; P < 0.00001 |
|
| In 35% of patients treated with intratympanic corticosteroid adverse events occurred including […] mild ear pain, severe ear pain (3 patients) | NR | 3/20 (15) severe ear pain | NR | NA | |
|
| Some patients had a tolerable pain reaction after the injection | NR | NR | NA | NA | |
| Other | Mood change | NR | 2/19 (10.5) | 8/20 (40) | 0.26 (0.06 to 1.08); favours IT corticosteroid; P = 0.06 | |
|
|
| Blood glucose problem | NR | 3/19 (15.8) | 6/20 (30) | 0.53 (0.15 to 1.81); favours IT corticosteroid; P = 0.31 |
|
|
| Sleep change | NR | 1/19 (5.3) | 6/20 (30) | 0.18 (0.02 to 1.32); favours IT corticosteroid; P = 0.09 |
|
|
| Increased appetite | NR | 1/19 (5.3) | 5/20 (25) | 0.21 (0.03 to 1.64); favours IT corticosteroid; P = 0.14 |
|
|
| Mouth dryness/thirst | NR | 0/19 (0) | 5/20 (25) | 0.10 (0.01 to 1.62); favours IT corticosteroid; P = 0.10 |
|
|
| Weight gain | NR | 0/19 (0) | 3/20 (15) | 0.15 (0.01 to 2.72); favours IT corticosteroid; P = 0.20 |
|
| No complications related to the treatment were noted in both the groups | NA | 0/25 (0) | 0/21 (0) | NA | |
|
| No long‐term complications were observed in any of the patients | NA | 0/19 (0) | 0/16 (0) | NA | |
|
| No side effects were observed in either group | NA | 0/32 (0) | 0/31 (0) | NA | |
|
| Apparent bleeding at intratympanic injection site | NR | 0/52 (0) | 0/52 (0) | NA | |
|
|
| External otitis or myringitis | NR | 0/52 (0) | 0/52 (0) | NA |
|
|
| Otitis media | NR | 0/52 (0) | 0/52 (0) | NA |
|
|
| Fluctuation of basal blood pressure (> 10 mmHg) | NR | 2/52 (3.8) | 7/52 (13.5) | NA |
|
|
| Fluctuation of fasting blood glucose (> 2 mmol/L) | NR | 5/52 (9.6) | 12/52 (23.1) | NA |
|
|
| Emotional change | NR | 8/52 (15.4) | 15/52 (28.8) | NA |
|
|
| Appetite change | NR | 13/52 (25.0) | 25/52 (48.1) | NA |
|
|
| Dyssomnia | NR | 23/52 (44.2) | 38/52 (73.1) | NA |
|
|
| Water‐sodium retention | NR | 9/52 (17.3) | 24/52 (46.1) | NA |
|
|
| Acne on face and body | NR | 2/52 (3.8) | 6/52 (11.5) | NA |
|
|
| Irregular menstruation | NR | 5/21 (23.8) | 11/23 (47.8) | NA |
|
|
| Cushing's syndrome | NR | 0/52 (0) | 1/52 (1.9) | NA |
|
|
| Osteoporotic fracture |
| 0 | 0 |
|
|
| In one case an acute suppurative otitis media developed that was eliminated by local antibacterial therapy. This patient was excluded from the study. | NA | NA | NA | NA | |
|
|
| Nine patients in the ST (standard therapy) group and 12 patients in the intravenous corticosteroid group complained of sleep loss | Completely corrected after withdrawal | NR | 9/24 (37.5) in ST group and 12/25 (48) in intravenous corticosteroid group | NA |
|
|
| No systemic adverse effects related to intratympanic application of steroids were noticed | NA | 0/24 (0) | NA | NA |
|
|
| No serious side effects related to systemic administration of steroids were observed in the study | NA | NA | 0/49 (0) | NA |
|
| No complications were seen in patients (unclear which group), including those with hypertension or diabetes | NA | NR | NR | NA | |
|
| Mood change | By the 6‐month follow‐up most adverse events had resolved | 12/129 (9.3) | 54/121 (44.6) | 0.21 (0.12 to 0.37); favours IT corticosteroid; P < 0.00001 | |
|
|
| Blood glucose problem | By the 6‐month follow‐up most adverse events had resolved | 21/129 (16.3) | 36/121 (29.8) | 0.55 (0.34 to 0.88); favours IT corticosteroid; P = 0.01 |
|
|
| Sleep change | By the 6‐month follow‐up most adverse events had resolved | 9/129 (7) | 44/121 (36.4) | 0.19 (0.1 to 0.38); favours IT corticosteroid; P < 0.00001 |
|
|
| Appetite change | By the 6‐month follow‐up most adverse events had resolved | 6/129 (4.7) | 28/121 (23.1) | 0.2 (0.09 to 0.47); favours IT corticosteroid; P = 0.0002 |
|
|
| Dry mouth/thirst | By the 6‐month follow‐up most adverse events had resolved | 5/129 (3.9) | 30/121 (24.8) | 0.16 (0.06 to 0.39); favours IT corticosteroid; P < 0.0001 |
|
|
| Weight change | By the 6‐month follow‐up most adverse events had resolved | 7/129 (5.4) | 22/121 (18.2) | 0.3 (0.13 to 0.67); favours IT corticosteroid; P = 0.004 |
|
|
| Ear infection | By the 6‐month follow‐up most adverse events had resolved | 7/129 (5.4) | 2/121 (1.7) | 3.28 (0.7 to 15.49); favours systemic corticosteroid; P = 0.13 |
|
|
| Any adverse event: "Adverse events were reported by 87.6% (106 of 121) of participants in the oral group and 89.9% (116 of 129) in the intratympanic group." Note: it is unclear whether 'adverse events' refers to those already reported (and listed for this study in this table). | By the 6‐month follow‐up most adverse events had resolved | 116/129 (89.9) | 106/121 (87.6) | 1.03 (0.94 to 1.12); favours systemic corticosteroid; P = 0.56 |
|
|
| Serious adverse events: "In the intratympanic treatment group, these included osteomyelitis of the toe, leukemia, myocardial infarction, bladder cancer, chest pain due to possible endocarditis, and exacerbation of pre‐existing chronic obstructive pulmonary disease. In the oral treatment group, the serious adverse events were myocardial infarction, cerebral hemorrhage, hyponatremia, hospitalization for possible transient ischemic attack, and syncope. The case of hyponatremia arose from worsening of pre‐existent mild renal insufficiency in a patient with type 2 diabetes that was deemed study‐related." | NR | 6/129 (4.7) | 5/121 (4.1) | 1.13 (0.35 to 3.59); favours systemic corticosteroid; P = 0.84 |
|
| No adverse effects were reported in either study group during the study period | NA | NA | NA | NA | |
|
| Temporary adverse events: temporary adverse events in 22.7% of patients treated with oral prednisolone which included puffiness of face, ulcers in mouth, increased appetite, diarrhea [...] | NR | NR | 5/22 (22.7) | NA | |
|
|
| Adverse events: in 35% of patients treated with intratympanic methylprednisolone adverse events occurred including mild ear pain, severe ear pain (3 patients), ringing sensation in ear and dizziness | NR | 7/20 (35) | NR | NA |
|
| No otitis media […] observed in any of the individuals at their final visit | NA | 0/30 (0) | 0/30 (0) | NA | |
|
| No significant complications during the intratympanic injections or the follow‐up period | NA | 0/33 (0) | NA | NA | |
| *Patients in the comparator group for this trial also received intratympanic (IT) corticosteroid at a later time point in the trial, therefore complications of IT treatment are included here (Huang 2021). The intervention group received 24 days of IT corticosteroid and the comparator group received 12 days of systemic (intravenous corticosteroid) followed by 12 days of IT corticosteroid. Rate ratios are not presented as they are not applicable to the comparison of interest (IT compared to systemic as primary therapy). CI: confidence interval; IT: intratympanic; NA: not applicable; NR: not reported; RR: risk ratio; TM: tympanic membrane | ||||||
| Intratympanic plus systemic corticosteroids (combined therapy) versus systemic corticosteroids alone as primary therapy | ||||||
| Patient or population: sudden sensorineural hearing loss Settings: initial therapy Intervention: combination of intratympanic and systemic steroid therapy Comparison: systemic steroid therapy | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | No of participants | Relative effect | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Systemic therapy (assumed risk) | Combined therapy (corresponding risk) | |||||
| Change in hearing threshold determined by PTA Range 0 dB to 140 dB Negative values represent lowering and positive values represent raising of the hearing threshold. A lower hearing threshold represents hearing improvement.
| The mean change in PTA ranged across control groups from | The mean change in PTA in the intervention groups was on average ‐8.55 dB greater (‐4.61 greater to ‐12.48 greater) | 435 (6 studies) | MD ‐8.55 dB (95% CI ‐12.48 to ‐4.61) | ⊕⊕⊝⊝ low1 | The change in hearing threshold may be slightly increased in participants who receive combined therapy. However, it is unclear whether this increase would be noticeable to patients. |
| Proportion of patients whose hearing is improved | 579 per 1000a | 735 per 1000 (666 to 816) | 788 (10 studies) | RR 1.27 (95% CI 1.15 to 1.41) | ⊕⊝⊝⊝ | The evidence is very uncertain as to whether combined therapy changes the proportion of participants whose hearing is improved. |
| Final hearing threshold determined by PTA A lower value represents better hearing | The mean final PTA ranged across control groups from 39.1 dB to 59 dB | The mean final PTA in the intervention groups was on average 9.11 dB lower (1.67 lower to 16.56 lower) | 194 (3 studies) | MD ‐9.11 dB (95% CI ‐16.56 to ‐1.67) | ⊕⊝⊝⊝
| Combined therapy may result in slightly lower (more favourable) final hearing thresholds compared to systemic corticosteroids alone (as primary therapy) but the evidence is very uncertain, and it is not clear whether the change would be of importance to patients. |
| Adverse eventsb | Events in control group | Events in intervention group | No of Participants | Relative effect | Certainty of the evidence | Comments |
| Persistent tympanic membrane perforation | Comparison not applicablec | 5 studies reported a rate between 0% (0/85) and 5.5% (2/36) for those who received an intratympanic injection | 474 (5 studies) | Not calculable | ⊕⊝⊝⊝ very low4 | The evidence is very uncertain regarding the risk of tympanic membrane perforation for those who received intratympanic steroids. |
| Vertigo/dizziness: timing not reported | No study reported on this outcome for both the intervention and comparator groups. | |||||
| Vertigo/dizziness: at the time of injection | Comparison not applicablec | 4 studies reported a rate between 0% (0/60) and 8.1% (3/37) for those who received an intratympanic injectiond | 341 (4 studies) | Not calculable | ⊕⊝⊝⊝ very low5 | The evidence is very uncertain regarding the risk of vertigo/dizziness at the time of intratympanic injection for those who received intratympanic corticosteroid as primary treatment. |
| Ear pain: timing not reported | No study reported on this outcome for both the intervention and comparator groups. | |||||
| Ear pain: at the time of injection | Comparison not applicablec | One study reported a rate of 5/37 (13.5%) | 73 (1 study) | Not calculable | ⊕⊝⊝⊝ very low6 | The evidence is very uncertain regarding the risk of ear pain at the time of intratympanic injection for those who received combined treatment as primary treatment. |
| *The basis for the assumed risk is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| aTen studies recruited participants suffering from sudden sensorineural hearing loss. The incidence of improvement for the 10 studies was 57.86%. We have used 579 per 1000 to express the assumed risk. bOnly the most widely reported adverse events are described here. For adverse events that could feasibly occur in either group, we have only included the studies that provided a rate for both groups. For adverse events that could only occur in one group, we have only included the studies that reported the rate in that group, and presented these as a range. A full description of adverse event data is available for reference in Table 2. cComparisons between patients receiving intratympanic therapy and those receiving only systemic therapy were regarded as invalid for the following adverse events: persistent tympanic membrane perforation, vertigo observed at the time of intratympanic injection and ear pain observed at the time of intratympanic injection. This is explained in Data extraction and management. dIn one study, two groups received intratympanic injection: one group received intratympanic corticosteroid and the other received intratympanic and systemic corticosteroid (Tsounis 2018).
GRADE Working Group grades of evidence | ||||||
| 1Downgraded one level due to risk of bias: we rated a study contributing moderate weight to the overall effect estimate as high risk of bias due to concern about random sequence generation and allocation concealment. Five studies were at high risk of other bias, and one study was at risk of attrition bias. Downgraded one level due to imprecision: the 95% CI overlaps the threshold for clinical relevance, taken to be 10 dB. 2Downgraded one level due to risk of bias: we judged 8 of 10 studies to be at high or unclear risk of selection bias and at high risk of other bias. Downgraded one level due to imprecision: the 95% CI overlaps the threshold for clinical relevance. Downgraded one level due to inconsistency: the I2 value was moderate (47%). 3Downgraded two levels due to risk of bias: we judged all three studies to be at high or unclear risk of selection bias and high risk of other bias. We also judged one of three studies to be at high risk of bias for incomplete outcome data and selective reporting. Downgraded one level due to imprecision: the 95% CI overlaps the threshold for clinical relevance and the sample size is smaller than the optimal information size (taken as 400 participants). 4Downgraded two levels due to risk of bias: we judged two studies to be at high risk of bias because of concern about random sequence generation, two studies because of selective reporting, one study because of concern about blinding and one study because of concern about allocation concealment. 5Downgraded two levels due to risk of bias: we judged one study to be at high risk of bias because of concern about random sequence generation and blinding, one study because of selective reporting and one study because of incomplete outcome data. Downgraded two levels due to imprecision: the number of events is smaller than the optimal information size (taken as 300 events) and an effect estimate could not be calculated. 6Downgraded two levels due to imprecision: the number of events is smaller than the optimal information size (taken as 300 events) and an effect estimate could not be calculated. Downgraded one level because of indirectness: provision of data by only a single study from a single setting, which may not adequately represent all patients with ISSNHL. | ||||||
| Adverse event reported | Study | How reported | Details of recovery | Rate in intervention group (%) | Rate in comparator group (%) | RR (95% CI) |
|---|---|---|---|---|---|---|
| TM perforation | No significant complications during or after IT dexamethasone, including TM perforation […] | NR | 0/60 (0) | NA | NA | |
|
| Two patients developed tympanic perforation (reported as 2.6% of whole study sample; unclear how many from each group) | Treated with cauterisation and paper patch (n = 1) and tympanoplasty (n = 1) | NR | NR | NA | |
|
| None of the patients had an important complication, namely […] TM perforation (unclear which group) | NR | NR | NR | NA | |
|
| No significant complications during or after IT steroid injection including TM perforation | NR | 0/19 (0) | NA | NA | |
|
| No case of residual TM perforation […] was noted | No long‐term complications resulting from either oral steroid or IT steroid in any of the patients | 0/37 (0) | NA | NA | |
| Vertigo/dizziness | No significant complications during or after IT dexamethasone, including […] vertigo […] | NR | 0/60 (0) | NA | NA | |
|
| None of the patients had an important complication, namely, […] vertigo […] (unclear which group) | NR | NR | NR | NA | |
|
| No significant complications during or after IT steroid injection, including […] vertigo […] | NR | 0/19 (0) | NA | NA | |
|
| Three patients complained of vertigo immediately after injection | Recovered after 2 hours of rest | 3/37 (8.1) | NA | NA | |
|
| One patient experienced transient dizziness as a result of caloric stimulation from the injected steroid solution (unclear which of two groups receiving intratympanic injection) | Symptoms resolved completely within 15 minutes and there was no need to discontinue the treatment. The injections that followed caused no further side effects. | NR | NR | NA | |
| Tinnitus | No significant | NR | 0/60 (0) | NA | NA | |
|
| None of the patients had an important complication, namely, […] tinnitus […] (unclear which group) | NR | NR | NR | NA | |
|
| No significant complications during or after IT steroid injection, including […] tinnitus […] | NR | 0/19 (0) | NA | NA | |
| Ear pain | None of the patients had an important complication, namely, […] otalgia […] (unclear which group) | NR | NR | NR | NA | |
|
| Otalgia occurred in 5 patients after IT corticosteroid injection | Relieved after 1 hour | 5/37 (13.5) | NA | NA | |
| Other | No significant complications during or after IT dexamethasone, including […] otitis media […] | NR | 0/60 (0) | NA | NA | |
|
| None of the patients had an important complication, namely, […] nystagmus, otitis media […] (unclear which group) | NR | NR | NR | NA | |
|
| No long‐term complications resulted from either the prednisone taper or the IT corticosteroid in any of the patients enrolled in the study | NR | 0/16 (0) | 0/18 (0) | NA | |
|
| No significant complications during or after IT steroid injection, including […] otitis media […] | NR | 0/19 (0) | NA | NA | |
|
| No case of […] otitis media was noted | No long‐term complications resulted from either oral steroid or IT steroid in any of the patients | 0/37 | NA | NA | |
|
|
| No long‐ term complications resulted from either oral steroid or IT steroid in any of the patients | NA | 0/37 (0) | 0/36 (0) | NA |
|
| No significant complications during the intratympanic injections or the follow‐up period | NA | 0/33 (0) | NA | NA | |
|
| No significant complications occurred during IT injections or the follow‐up period. One case of otitis media was encountered (unclear which group) | NR | 0/46 (0) (significant complications) | NA | NA | |
| IT: intratympanic; NA: not applicable; NR: not reported; TM: tympanic membrane | ||||||
| Intratympanic corticosteroids versus no treatment or versus placebo as secondary therapy | ||||||
| Patient or population: sudden sensorineural hearing loss Settings: after treatment failure with systemic steroids Intervention: intratympanic steroid therapy Comparison: no treatment/placebo | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | No of participants | Relative effect | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| No treatment/placebo (assumed risk) | Intratympanic therapy (corresponding risk) | |||||
| Change in hearing threshold determined by PTA Range 0 dB to 140 dB Negative values represent lowering and positive values represent raising of the hearing threshold. A lower hearing threshold represents hearing improvement.
| The mean change in PTA ranged across control groups from | The mean change in PTA in the intervention groups was on average ‐9.07 dB greater (‐6.66 greater to ‐11.47 greater) | 280 (7 studies) | MD ‐9.07 dB (95% CI ‐11.47 to ‐6.66) | ⊕⊕⊝⊝
| Intratympanic therapy may have a small effect on hearing threshold compared to no treatment or placebo (as secondary therapy), but it is not clear whether this change would be important to patients. |
| Proportion of patients whose hearing is improved | 70 per 1000a | 385 per 1000 (203 to 747) | 232 (6 studies) | RR 5.55 (95% CI 2.89 to 10.68) | ⊕⊕⊝⊝ | Intratympanic therapy may result in a much higher proportion of patients whose hearing is improved, compared to no treatment or placebo (as secondary therapy). |
| Final hearing threshold determined by PTA (a lower value represents better hearing) | The mean final PTA ranged across control groups from 59.9 to 90.5 dB HL | The mean final PTA in the intervention groups was on average ‐11.09 dB lower (‐4.72 lower to ‐17.46 lower) | 203 (5 studies) | MD ‐11.09 dB (95% CI ‐17.46 to ‐4.72) | ⊕⊕⊝⊝ | Intratympanic therapy may result in lower (more favourable) final hearing thresholds compared to no treatment or placebo (as secondary therapy). |
| Adverse eventsb | Events in control group | Events in intervention group | No of Participants (studies) | Relative effect (95% CI) | Certainty of the evidence (GRADE) | Comments |
| Persistent tympanic membrane perforation | Comparison not applicablec | 5 studies reported a rate between 0% (0/19) and 4.2% (1/24) for those who received an intratympanic injectiond | 185 (5 studies) | Not calculable | ⊕⊝⊝⊝ very low4 | The evidence is very uncertain regarding the risk of tympanic membrane perforation for those who received intratympanic injection (either corticosteroid or placebo) as secondary treatment. |
| Vertigo/dizziness: timing not reported | No study reported on this outcome for both the intervention and comparator groups. | |||||
| Vertigo/dizziness at the time of intratympanic injection | Comparison not applicablec | 3 studies reported a rate between 6.7% (1/15) and 33% (number not reported) for those who received an intratympanic injection.d | 128 (3 studies) | Not calculable | ⊕⊝⊝⊝ very low5 | The evidence is very uncertain regarding the risk of vertigo/dizziness at the time of intratympanic injection (either corticosteroid or placebo) as secondary treatment. |
| Ear pain: timing not reported | No study reported on this outcome for both the intervention and comparator groups. | |||||
| Ear pain at the time of intratympanic injection | Comparison not applicablec | One study reported no participants with ear pain at the time of intratympanic injection (0/24) | 44 (one study) | Not calculable | ⊕⊝⊝⊝ very low6 | The evidence is very uncertain regarding the risk of ear pain at the time of intratympanic corticosteroid injection as secondary treatment. |
| *The basis for the assumed risk is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| aSix studies recruited participants suffering from sudden sensorineural hearing loss after treatment failure with systemic steroids. The incidence of improvement for the control group in these six studies was 6.96%. We have used 70 per 1000 to express the assumed risk. bOnly the most widely reported adverse events are described here. For adverse events that could feasibly occur in either group, we have only included the studies that provided a rate for both groups. For adverse events that could only occur in one group, we have only included the studies that reported the rate in that group, and presented these as a range. A full description of adverse event data is available for reference in Table 3. cComparisons between patients receiving intratympanic therapy and those receiving only systemic therapy were regarded as invalid for the following adverse events: persistent tympanic membrane perforation, vertigo observed at the time of intratympanic injection and ear pain observed at the time of intratympanic injection. This is explained in Data extraction and management. dThis includes participants who received placebo intratympanic injection.
GRADE Working Group grades of evidence | ||||||
| 1Downgraded one level due to risk of bias: we rated one study contributing moderate weight to the overall effect estimate as having high risk of bias due to incomplete outcome data. All studies were at high risk of other bias. Downgraded one level due to imprecision: the 95% CI for the effect overlaps the threshold for clinical relevance and the sample size is smaller than the optimal information size (taken as 400 participants). One study included treatment in the comparator arm with vitamin B, vasodilators and benzodiazepines (Ho 2004). However, as the weight of this study in the meta‐analysis was low and exclusion of the study made little difference to the effect estimate we did not downgrade for indirectness. 2Downgraded one level due to risk of bias: we rated two studies as being at high risk of bias due to selective reporting and one study was at high risk of bias for incomplete outcome data. All studies were at high risk of other bias. Downgraded one level due to imprecision: the total number of events is smaller than the optimal information size (taken as 300 events). 3Downgraded one level due to risk of bias: we rated one study contributing moderate weight to the overall effect estimate as high risk of bias due to incomplete outcome data, and one other study as high risk of bias because of selective reporting. All studies were at high risk of other bias. Downgraded one level due to imprecision: the 95% CI for the effect overlaps the threshold for clinical relevance and the sample size is smaller than the optimal information size (taken as 400 participants). 4Downgraded one level due to risk of bias: we rated one study as high risk of bias because of selective reporting; we rated one study as high risk of bias because of incomplete outcome data; we rated three studies as uncertain for random sequence generation, allocation concealment and blinding. Downgraded two levels because of imprecision: the number of events is smaller than the optimal information size (taken as 300 events) and an effect estimate could not be calculated. 5Downgraded one level due to risk of bias: we rated one study as high risk of bias because of selective reporting; we rated one study as high risk of bias because of incomplete outcome data; we rated two studies as uncertain for random sequence generation, allocation concealment and blinding. Downgraded two levels because of imprecision: the number of events is smaller than the optimal information size (taken as 300 events) and an effect estimate could not be calculated. 6Downgraded two levels because of imprecision: the number of events is smaller than the optimal information size (taken as 300 events) and an effect estimate could not be calculated. Downgraded one level due to indirectness: single study from a single setting, which may not adequately represent all patients with ISSNHL. | ||||||
| Adverse event reported | Study | How reported | Details of recovery | Rate in intervention group (%) | Rate in comparator group (%) | RR (95% CI) |
|---|---|---|---|---|---|---|
| TM perforation | No residual TM perforation | NA | 0/15 (0) | NA | NA | |
|
| Persistent TM perforation | No hearing loss in the affected ear. The perforation was treated successfully with a paper patch. | 1/24 (4.2) | NA | NA | |
|
| One patient (unclear which group) had a major catheter dislocation with perforation of ear drum. Note: both groups received IT injection, either corticosteroid or normal saline. | Small ear drum perforation was closed with a myringoplasty | NR | NR | NA | |
|
| Transient TM perforation | Healed spontaneously by follow‐up 1 month later | 1/27 (3.7) | NR | NA | |
|
| No TM perforation was noticed at last visit | NA | 0/19 (0) | NA | NA | |
| Vertigo/dizziness | Complained of vertigo immediately after injection | Recovered after 2 hours of rest | 1/15 (6.7) | NA | NA | |
|
| Three patients complained of vertigo [...] during the injections | Resolved within minutes | 3/24 (12.5) | NA | NA | |
|
|
| No disequilibrium | NA | 0/24 (0) | NR | NA |
|
|
| No dizziness for more than 24 hours | NA | 0/24 (0) | NR | NA |
|
| One patient (unclear which group) with increase in vertigo | Resolved | NR | NR | NA | |
|
| Temporary dizziness experienced by one‐third of subjects (unclear how many each group). Note: both groups received IT injection, either corticosteroid or normal saline. | Relieved by resting for a short time. Three participants quit the trial because of uncomfortable dizziness (unclear how many each group). | NR | NR | NA | |
| Tinnitus | Three patients complained of vertigo or an increase in tinnitus during the injections | Resolved within minutes | 3/24 (12.5) | NA | NA | |
| Hearing loss | The injection did not induce an increase in […] hearing loss […] for greater than 24 h | NA | 0/24 (0) | NR | NA | |
|
| No participant experienced a decrease in hearing of 10 dB or more | NA | 0/27 | 0/28 | NA | |
| Ear pain | The injection did not induce an increase in ear pain […] | NA | 0/24 (0) | NA | NA | |
|
| Two patients (unclear how many each group) with ear pain. Note: both groups received IT injection, either corticosteroid or normal saline. | Resolved | NR | NR | NA | |
|
| A mild ear pain occurring the first hour post‐injection | Easily controlled with common analgesics | NR | NA | NA | |
| Other | One of 15 patients had acne | NR | 1/15 (6.7) | NR | NA | |
|
| No serious complications such as chronic otitis media, disequilibrium or dysgeusia developed | NA | 0/24 (0) | NR | NA | |
|
| One patient (unclear which group) with each of: ear canal skin defect, steroid acne, nausea after antibiotic intake, gastroenteritis, hypokalaemia, pump battery failure and viral conjunctivitis. Three patients with headache (unclear how many in each group; one considered as 'possibly', 'probably' or 'very likely' related to the study) and 3 (unclear how many each group) with increased liver function tests (probably due to antibiotics). | Resolved | NR | NR | NA | |
|
| Severe nausea or vomiting was not experienced by any of the participants after the injection therapy | NA | 0/27 | 0/28 | NA | |
|
| No infection was noticed in any of the patients at their last visit | NA | 0/19 (0) | 0/18 (0) | NA | |
| TM: tympanic membrane; NA: not applicable; NR: not reported; IT: intratympanic | ||||||
| Intratympanic plus systemic corticosteroids (combined therapy) versus systemic corticosteroids alone as secondary therapy | ||||||
| Patient or population: sudden sensorineural hearing loss Settings: after treatment failure with systemic steroids Intervention: combination of intratympanic and systemic steroid therapy Comparison: systemic steroid therapy | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | No of participants | Relative effect | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Systemic therapy (assumed risk) | Combined therapy (corresponding risk) | |||||
| Change in hearing threshold determined by PTA | No studies reported this outcome. | |||||
| Proportion of patients whose hearing is improved | 205 per 1000a | 459 per 1000 (226 to 933) | 76 (1 study) | RR 2.24 (95% CI 1.10 to 4.55) | ⊕⊝⊝⊝ | Combined therapy may increase the proportion of patients whose hearing is improved compared to systemic corticosteroids alone (as secondary therapy), but the evidence is very uncertain. |
| Final hearing threshold determined by PTA | No studies reported this outcome. | |||||
| Adverse eventsb | Events in control group | Events in intervention group | No of participants | Relative effect | Certainty of the evidence | Comments |
| Persistent tympanic membrane perforation | Comparison not appropriatec | One study reported a rate of 8.1% (3/37) | 76 (1 study) | Not calculable | ⊕⊝⊝⊝ very low2 | The risk of tympanic membrane perforation among those who receive intratympanic corticosteroid combined with systemic corticosteroid as primary treatment is very uncertain. |
| Vertigo/dizziness: timing not reported | No studies reported this outcome. | |||||
| Vertigo/dizziness: at the time of injection | Comparison not appropriatec | No study reported a rate in the intervention group. | ||||
| Ear pain: timing not reported | No studies reported this outcome. | |||||
| Ear pain: at the time of injection | Comparison not appropriatec | No study reported a rate in the intervention group. | ||||
| *The basis for the assumed risk is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| aOne study recruited participants suffering from sudden sensorineural hearing loss after treatment failure with systemic steroids. The incidence of improvements was 20.51%. We have used 205 per 1000 to express the assumed risk. bOnly the most widely reported adverse events are described here. For adverse events that could feasibly occur in either group, we have only included the studies that provided a rate for both groups. For adverse events that could only occur in one group, we have only included the studies that reported the rate in that group, and presented these as a range. A full description of adverse event data is available for reference in Table 4. cComparisons between patients receiving intratympanic therapy and those receiving only systemic therapy were regarded as invalid for the following adverse events: persistent tympanic membrane perforation, vertigo observed at the time of intratympanic injection and ear pain observed at the time of intratympanic injection. This is explained in Data extraction and management.
GRADE Working Group grades of evidence | ||||||
| 1Downgraded two levels due to risk of bias: we judged the study to be at high risk of selection bias, performance bias, incomplete outcome data, selective reporting and other bias. Downgraded two levels due to imprecision: the 95% CI overlaps the threshold for clinical relevance and the total number of events is smaller than the optimal information size (taken as 300 events). 2Downgraded two levels due to risk of bias: we judged the study to be at high risk of bias because of selection bias, concern about blinding, incomplete outcome data and selective reporting. Downgraded two levels due to imprecision: the number of events is smaller than the optimal information size (taken as 300 events) and an effect estimate could not be calculated. Downgraded one level because of indirectness: provision of data by only a single study from a single setting, which may not adequately represent all patients with ISSNHL. | ||||||
| Adverse event reported | Study | How reported | Details of recovery | Rate in intervention group (%) | Rate in comparator group (%) | RR (95% CI) |
|---|---|---|---|---|---|---|
| TM perforation | Three patients had small eardrum perforations | Successful closure by simple treatment | 3/37 (8.1) | NA | NA | |
| Vertigo | Second frequent complaint: transient vertigo after the drug had been injected into the ear | Not a severe problem if the drug was heated in 37°C water before injection and the vertigo disappeared after a few minutes or under 30 minutes | NR | NA | NA | |
| Ear pain | Most frequent complaint | Easily controlled by the oral administration of paracetamol 30 minutes before the local infusion of the methylprednisolone | NR | NA | NA | |
| Hearing loss* | No loss in hearing related to the treatment (in either group) | NA | 0/37 (0) | 0/39 (0) | NA | |
| Other | One patient had tongue paresthesia (unclear which group) | Resolved after 2 weeks | NR | NR | NA | |
|
|
| No infections were observed (unclear which group) | NA | NR | NR | NA |
|
|
| Long‐term complications did not occur in any patients who received the transtympanic injections | NA | 0/37 (0) | NR | NA |
| *Hearing loss defined as ≥ 15 dB worsening in pure tone audiometry or ≥ 15% worsening of speech discrimination score. NA: not applicable; NR: not reported; TM: tympanic membrane | ||||||
Background
Description of the condition
Idiopathic sudden sensorineural hearing loss (ISSNHL) is a sudden decrease in sensorineural hearing sensitivity of unknown aetiology. It is usually unilateral and the degree of severity can vary from mild hearing loss to total deafness. It may also be accompanied by vertigo and tinnitus.
There is no international consensus on the definition of ISSNHL in terms of the degree of hearing threshold change or the number of specific frequencies that are affected on pure tone audiological testing. A definition that is commonly used is "loss of at least 30 dB in three connected frequencies within 72 hours" (Chandrasekhar 2019; NIDCD 2018). However, this definition is not universally accepted. It does not specify the frequencies and the frequency range, the rational for choosing this threshold is not known and it is often not used as an inclusion criterion in clinical trials on ISSNHL. Although for mild and moderate hearing losses, statistical floor effects complicate the evaluation of recovery (Chen 2003), it appears justified to expand the definition to cases with less than 30 dB of hearing loss (Chandrasekhar 2019; Plontke 2007). There is also a lack of consensus on the most appropriate outcome criteria for clinical studies (Plontke 2007).
The incidence of sudden sensorineural hearing loss has been estimated to be 5 to 20 per 100,000 per year in industrial countries (Byl 1977; Hughes 1996; Stokroos 1996). However, according to studies in Germany, the incidence may be much higher: Olzowy 2005 estimated the incidence at 160 per 100,000 per year and Klemm 2009 estimated it at 400 per 100,000 per year. This discrepancy may be due to the absence of international consensus on the audiological definition and outcome criteria. The mean age of patients included in randomised controlled trials (RCTs) is between 45 and 55. Men and women are equally affected. ISSNHL in childhood is rare (Desloovere 1988; Klemm 2007; Mösges 2009; Plontke 2007; Probst 1992; Tucci 2002; Tran Ba Huy 2005). Idiopathic sudden sensorineural hearing loss, particularly when accompanied by tinnitus and dizziness, results in a significant reduction in quality of life (Carlsson 2011; Stachler 2012).
Various theories to explain ISSNHL have been proposed, for example viral infection, vascular occlusion, breakdown of labyrinthine membranes or barriers, immune‐mediated mechanisms (Vambutas 2021) and abnormal cellular stress responses within the cochlea. However, none of these hypotheses has been proven convincingly in humans (Merchant 2005; Merchant 2008).
Treatment modalities for ISSNHL are mostly based on the above etiopathogenetic hypotheses and include (gluco)corticosteroids, rheological drugs (e.g. dextran, hydroxyethyl starch, pentoxifylline and naftidrofuryl), vasodilators, anaesthetics, osmotically active substances, antioxidants, thrombocyte aggregation inhibitors, fibrinogen reduction through drugs or apheresis or rheopheresis (Suckfüll 2002), hyperbaric oxygen therapy, antiviral therapy, N‐methyl‐D‐aspartate (NMDA) receptor antagonists, immunosuppressants, anti‐apoptotic substances (Suckfuell 2014), and other substances (see reviews in: Conlin 2007a; Labus 2010; Lawrence 2015; Plontke 2005). Cochrane Reviews have assessed treatment of ISSNHL with systemic corticosteroids (Wei 2006; Wei 2013), hyperbaric oxygen (Bennett 2007; Bennett 2012), and vasodilators (Agarwal 2009), without demonstrating clear efficacy.
Systemic corticosteroids are widely used for ISSNHL worldwide (Plontke 2005). A Cochrane Review on systemic corticosteroids for ISSNHL found that there was uncertainty about the value of corticosteroids in the treatment of ISSNHL, "since the evidence from randomised controlled trials is contradictory in outcome, in part because the studies are based upon too small a number of patients" (Wei 2006). These findings were supported by another meta‐analysis (Conlin 2007b). The updated version of the Cochrane Review also included a randomised, placebo‐controlled, multicentre trial published in 2012 comparing the effect of prednisolone and placebo (Nosrati‐Zarenoe 2012); again the review found that there was uncertainty about the value of systemic corticosteroids in the treatment ISSNHL (Wei 2013).
In general, possible side effects of systemic corticosteroid medication include metabolic complications, such as glucose intolerance and diabetes mellitus, hypertension, increased intraocular pressure and glaucoma, psychotropic effects, hypothalamic‐pituitary‐adrenal‐axis suppression, gastrointestinal bleeding, bone loss, avascular necrosis of the femoral or humeral head and potential infections. A study investigating the risk of corticosteroid‐induced hyperglycaemia concluded that prevalence during systemic therapy is high and rises as the dose increases (Rohrmeier 2012). Although the rate of occurrence of side effects with systemic corticosteroid therapy appears low (Garcia‐Berrocal 2008), systematic data recording and publication of the proposed side effects are still insufficient, and adverse effects from a short course of high‐dose systemic corticosteroids have not been documented with good evidence. It is only possible, therefore, to speculate as to whether these known side effects occur during (longer) systemic corticosteroid treatment of ISSNHL and, if so, to what degree.
The terms 'steroids', 'corticosteroids', 'glucocorticoids' are unfortunately used imprecisely and interchangeably in the literature on ISSNHL. The term 'corticosteroid' is used throughout this review, since this term is more often used and generally accepted in the literature on ISSNHL (Chandrasekhar 2019; Rauch 2011).
Description of the intervention
The rationale for local intratympanic application of drugs for the treatment of inner ear diseases is based on the expected advantages over systemic treatment. These are 1) the bypassing of the blood‐labyrinthine barrier, resulting in 2) higher concentrations in the inner ear fluids despite the lower total amount of drug given, and 3) avoiding the major unwanted effects of systemically administered medications due to lower systemic drug levels.
Pharmacokinetic studies in animals and humans have shown that high doses of systemic corticosteroids are needed to achieve detectable drug levels in the inner ear perilymph and that substances applied to the round window membrane lead to significantly higher drug levels in the inner ear fluids compared to systemic application (Bachmann 2001; Bird 2007; Bird 2011; Chandrasekhar 2000; Niedermeyer 2003; Parnes 1999). Thus, applying drugs locally may be more effective in treating sudden sensorineural hearing loss and may avoid systemic complications and side effects. The introduction of this drug delivery approach has triggered a large number of pre‐clinical studies focused on the pharmacokinetics of local drug delivery to the inner ear and the development of drug delivery systems (reviewed, for example, in: Hoskison 2013; Nakagawa 2011; Pararas 2012; Salt 2009; Salt 2018).
Intratympanic injection of corticosteroids for ISSNHL in humans was pioneered by Silverstein (Silverstein 1996) and Parnes (Parnes 1999). Since then, a rapidly growing number of reports have been published on treatment results of intratympanic application of corticosteroids for inner ear disorders (Lavigne 2016; Liebau 2017; Liebau 2018). Intratympanic injection of corticosteroids is used not only as a single treatment approach but also in combination with systemic administration of corticosteroids. In a Cochrane meta‐analysis, Phillips 2011 assessed the efficacy of intratympanic corticosteroids for Ménière's disease. The majority of clinical reports, however, described the use of intratympanic corticosteroids for sudden hearing loss and more studies, including randomised controlled trials, are ongoing. So far, mainly dexamethasone or methylprednisolone preparations have been used as a primary or a second‐line ('rescue', 'salvage', 'reserve') intratympanic therapy for ISSNHL. Although these studies have shown intratympanic treatment with corticosteroids to be relatively safe, efficacy is difficult to assess since many studies did not compared their findings with a control group, and an even smaller number were randomised trials (reviewed in: Chandrasekhar 2019; Crane 2015; Gao 2016; Garavello 2012; Haynes 2007; Lavigne 2016; Lawrence 2015; Li 2015; Marx 2018 Ng 2015; Seggas 2011; Spear 2011; Stachler 2012; Vlastarakos 2012; Zhao 2016).
Several methods for intratympanic application of corticosteroids haven been developed in recent years. Most are single or repeated intratympanic injections with or without visualisation of the round window membrane. In some studies additional substances like hyaluronic acid are used for volume stabilisation to increase the persistence of the drug in the middle ear. Another strategy is continuous or discontinuous drug application via partly or fully implantable pump systems, allowing adjustment of drug concentrations over time. Potential adverse events in the intratympanic application of corticosteroids are in principle the same as in systemic drug administration, but to a lesser extent. Some of the intratympanically applied drug may be lost from the middle ear by drainage through the Eustachian tube and then swallowed. However, the dose is much lower than with systemic application. Persistent perforation of the tympanic membrane can develop after injection if there is an impaired healing process. Also, temporary pain may be observed, or temporary vertigo or dizziness due to caloric stimulation.
How the intervention might work
Corticosteroids were originally implemented in the treatment of ISSNHL because of their anti‐inflammatory effect. It is assumed that the main cause of sudden deafness is a harmful effect of the immune system on the inner ear in response to viral infection (Wilson 1980). However, corticosteroids have further effects, mainly mediated by activation of the glucocorticoid receptor, which could play a role in the treatment of sudden hearing loss. One important effect is an increase in anti‐apoptotic transcription factors in cells and the blocking of apoptosis signalling pathways. This could protect the sensory hair cells and other neural and non‐neural structures in the inner ear (Eshraghi 2006; Hoang 2009; Trune 2012). However, probably the most important effect of corticosteroids is their property of reducing the impact of oxidative stress in cells (Trune 2012). Recent studies point out that oxidative stress plays an important role in the aetiology of sudden hearing loss (Gul 2016; Quaranta 2016). Glucocorticoids also bind to the mineralocorticoid receptor (Grossmann 2004). Additional effects of corticosteroids are mediated by activation of the mineralocorticoid receptor, which has an impact on cochlear ion transport (Trune 2006). This may help to restore a disturbed homeostasis in the inner ear and ensure hair cell function driven by the endocochlear potential (MacArthur 2015). For some of these effects, especially the anti‐apoptotic and anti‐oxidative effects, a high drug concentration in the inner ear might be necessary (Haake 2009). Since there is no accumulation of corticosteroids in the inner ear, and as corticoid entrance to the inner ear structures is limited by the blood‐labyrinth barrier, drug concentration in the inner ear with systemic application will not exceed the systemic plasma concentration. Local application of corticosteroids to the inner ear by intratympanic injection achieves a short‐duration, high concentration of the drug in the middle ear cavity, from where the drug can diffuse through the boundaries to the inner ear, i.e. the round window and the oval window (King 2011; Salt 2009). Thus, higher corticosteroid concentrations can be achieved in the inner ear, which might be necessary for successful treatment of sudden hearing loss (Bird 2007; Bird 2011).
Hearing recovery in patients with ISSNHL mostly occurs early, within a few days of onset, but can also occur after several weeks (Liebau 2017; Liebau 2018). It has been observed that the time course of hearing recovery can extend to six months (Kosyakov 2011). However, most of the hearing improvement will take place during the first weeks after onset. To estimate the treatment effect of an intervention for ISSNHL, it is desirable that the final outcome assessment is conducted after several weeks of follow‐up. Short evaluation periods may risk underestimation of treatment effects (Wycherly 2011). After the recovery period, the hearing thresholds reached can be assumed to be stable in most patients. A large randomised controlled trial (published protocol) assessed the primary outcome 30 days after onset with an initial assessment after 10 days and a follow‐up at six months (Plontke 2016).
Why it is important to do this review
There is still uncertainty as to 1) whether intratympanic corticosteroids are better than placebo or no treatment, 2) whether intratympanic administration of corticosteroids alone or in combination with systemic application of steroids will lead to better results than systemic drug administration alone, 3) if so, which treatment protocol will lead to the best outcome and 4) what risks of adverse events are associated with this approach in inner ear therapy. This Cochrane Review was therefore warranted to assess the benefits and harms of intratympanic corticosteroids treatment for ISSNHL.
Objectives
To assess the effects of intratympanic corticosteroids in people with idiopathic sudden sensorineural hearing loss (ISSNHL).
Methods
Criteria for considering studies for this review
Types of studies
We included randomised controlled trials (RCTs) and quasi‐randomised controlled trials according to the Cochrane definition (Handbook 2021). Cross‐over trials were not included. Cross‐over trials are not feasible for the evaluation of interventions in the treatment of ISSNHL as there is no possibility to return to the baseline situation after the first intervention.
Types of participants
We included adults and children, female and male, of any ethnic origin, with unilateral ISSNHL (i.e. sudden sensorineural hearing loss of unknown aetiology) with or without vertigo, and with or without tinnitus.
Studies in patients with non‐idiopathic conditions or diagnoses were excluded (e.g. acoustic trauma, Ménière's disease, fluctuating hearing loss, endolymphatic hydrops, suspected retro‐cochlear lesion, hearing loss due to ear surgery, perilymph fistula or barotrauma, middle ear inflammation or effusion, or conductive hearing loss).
Types of interventions
Corticosteroids (also referred to as steroids), which were applied by intratympanic application for the treatment of ISSNHL as one of two treatment strategies:
-
as primary (first‐line) treatment; or
-
as secondary (rescue/salvage/reserve/second‐line) treatment after failure of primary therapy.
Corticosteroids were administered using one of the following drug delivery systems:
-
single or repeated intratympanic injection with or without volume stabilisation and with or without visualisation of the round window membrane; or
-
continuous or discontinuous drug application via partly or fully implantable pump systems.
The different methods of intratympanic drug delivery were considered together as intratympanic application.
We included studies of the following comparisons:
-
intratympanic corticosteroids versus no treatment or versus placebo;
-
intratympanic corticosteroids versus systemic corticosteroids;
-
intratympanic plus systemic corticosteroids (combined therapy) versus systemic corticosteroids alone;
-
intratympanic plus systemic corticosteroids (combined therapy) versus no treatment or versus placebo.
Studies were included regardless of the precise details of the treatment protocol (e.g. type of corticosteroid used, injection procedure, dose, frequency of application and duration of treatment).
Types of outcome measures
We did not use the outcomes selected for the review as a basis for including or excluding studies. We conducted analyses on outcome data collected more than one week (eight days or more) after the start of treatment.
The primary outcome was the change in mean hearing threshold determined by pure tone audiometry (pure tone average) between treatment arms and measured in decibels (dB). A lowering of the mean hearing threshold represents an improvement in hearing. To indicate the direction of change, we denoted a lowering of mean threshold as a negative value, and an elevation of the mean threshold as a positive value. There was no restriction on frequencies or number of frequencies used for generation the pure tone average.
Secondary outcome measures included final hearing threshold (pure tone average at the study endpoint), frequency‐specific changes in mean hearing threshold, the proportion of patients whose hearing improved (based on pure tone average and/or speech audiometry and without restriction on definition) and changes in hearing threshold based on speech audiometry (without restriction on type or language of speech test).
Also among the secondary outcomes were minor and serious adverse events.
Primary outcomes
-
Change in hearing threshold with pure tone audiometry (pure tone average (PTA)).
Secondary outcomes
-
Proportion of patients whose hearing is improved (criteria for improvement were defined by the included studies).
-
Final hearing threshold with pure tone audiometry.
-
Change in hearing threshold with speech audiometry.
-
Frequency‐specific changes in hearing threshold with pure tone audiometry.
-
Mean level of improvement in those whose hearing is improved.
-
For patients with profound pre‐treatment hearing loss: percentage of patients reaching serviceable hearing (defined as maximum percentage of correctly understood monosyllables equal or greater than 50%).
-
Effect on tinnitus and vertigo.
-
Minor and serious adverse events.
Search methods for identification of studies
The Cochrane ENT Information Specialist conducted systematic searches for randomised controlled trials and controlled clinical trials. There were no language, publication year or publication status restrictions. The date of the search was 23 September 2021.
Electronic searches
The Information Specialist searched:
-
the Cochrane ENT Trials Register (searched via the Cochrane Register of Studies 23 September 2021);
-
the Cochrane Central Register of Controlled Trials (searched via the Cochrane Register of Studies) (CENTRAL 2021, Issue 9);
-
PubMed (1946 to 23 September 2021);
-
Ovid Embase (1974 to 23 September 2021);
-
LILACS, lilacs.bvsalud.org (searched 23 September 2021);
-
Web of Knowledge, Web of Science (1945 to 23 September 2021);
-
CNKI, www.cnki.com.cn (searched via Google Scholar 23 September 2021);
-
ClinicalTrials.gov (searched via the Cochrane Register of Studies and clinicaltrials.gov 23 September 2021);
-
World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP), www.who.int/ictrp (searched 23 September 2021).
The Information Specialist modelled subject strategies for databases on the search strategy designed for CENTRAL. Where appropriate, they were combined with subject strategy adaptations of the highly sensitive search strategy designed by Cochrane for identifying randomised controlled trials and controlled clinical trials (as described in the Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0, Box 6.4.b. (Handbook 2011). Search strategies for major databases including CENTRAL are provided in Appendix 1.
Searching other resources
We scanned the reference lists of identified publications for additional trials and contacted trial authors where necessary. In addition, the Information Specialist searched PubMed, the Cochrane Library and Google to retrieve existing systematic reviews relevant to this systematic review, so that we could scan their reference lists for additional trials.
Data collection and analysis
Data collection and analysis for this review were specified in a pre‐published protocol (Plontke 2009). Changes that have been made since the protocol are specified in the section Differences between protocol and review.
Selection of studies
After scanning all search results and independent screening of titles and abstracts, we retrieved the full texts of reports that loosely met the inclusion criteria and where exclusion of studies could not be clearly inferred from the abstract. At least two authors reviewed these and applied the inclusion criteria independently. These were 1) intratympanic corticosteroid treatment of ISSNHL, 2) clinical study, 3) stated randomisation process and 4) studying at least one comparison included in the review.
Final decisions on inclusion were based on full‐text analysis of preselected studies for the following criteria: 1) a reported randomisation process in the main text of the study report, 2) studying comparisons included in the review, 3) the diagnosis of included patients was ISSNHL, 4) the proportion of included patients with bilateral ISSNHL was below 5%, and 5) the time point of final outcome assessment was at least one week (eight days or more) after the start of treatment. In order to include a high number of studies, we also included those performing outcome assessments with short follow‐up (less than four weeks). However, we considered a short follow‐up duration of two weeks or less to represent a high risk of bias in these studies.
We openly discussed any differences of opinion about which studies to include in the review. If consensus could not be reached, we planned to refer these studies to the Cochrane ENT Co‐ordinating Editor. However, this was not necessary in any case. Publications in languages that could not be read by the authors were fully translated by a professional translator. Further, if such studies were included in the meta‐analyses, two native speakers independently performed extraction of key data, co‐ordinated by Cochrane ENT.
Data extraction and management
Study characteristics and data related to participants from each study were always independently extracted by at least three authors. Any discrepancies among the extracted data were discussed and resolved by consensus. Only treatment arms that met the comparisons defined in the review were included. If a study had more than one treatment arm matching the same type of intervention, we selected the one most widely used among included studies. In most studies, we extracted outcome data from the defined primary endpoint. When no time point was defined as the primary endpoint, we selected the latest time point for inclusion. Exceptionally, if the number of participants lost to follow‐up was very high at the final time point and the necessary outcome parameters and numbers of participants were reported for an earlier time point, we chose this earlier time point. This is documented in the Characteristics of included studies table.
We documented the following details for each study:
-
Methods (study design, country, year, setting, allocation, blinding).
-
Participants (inclusion and exclusion criteria, number of included participants, baseline parameters).
-
Interventions (treatment arms, time point of start of intervention, whether primary therapy or secondary/rescue therapy, dosage and type of steroid, drug delivery strategy, injection regime, duration of intervention, time to follow‐up, concomitant treatments).
-
Outcomes (defined primary and secondary outcomes in the review (see above), definition of PTA and successful hearing improvement, number of completed and analysed participants, type of analysis).
-
Funding sources and declarations of interest.
We always extracted outcome data based on intention‐to‐treat (ITT) analysis when they were reported, in preference to data based on per‐protocol (PP) analysis. We extracted the following summary statistics for each study and each outcome:
-
For continuous data: the mean values, standard deviations and number of patients for each treatment group.
-
For binary data: the number of participants experiencing an event and the number of patients assessed at the time point.
We took an exploratory approach to assessing adverse events and extracted data on all adverse events reported by the trialists.
Assessment of risk of bias in included studies
Assessment of the risk of bias of the included studies was undertaken independently by four authors with the following domains taken into consideration, as guided by the Cochrane Handbook for Systematic Reviews of Interventions (Handbook 2011):
-
sequence generation;
-
allocation concealment;
-
blinding;
-
incomplete outcome data;
-
selective reporting; and
-
other sources of bias.
We used the Cochrane risk of bias tool in RevMan 5 (RevMan 2020), which involves scrutiny of each domain as reported in the trial and judgement about the adequacy of each entry. Discrepancies between raters' judgements were discussed and resolved by consensus. We judged the risk of bias to be 'high', 'low' or 'unclear' and documented this together with an explanation in the risk of bias tables in Characteristics of included studies. In non‐placebo‐controlled studies, we generally considered the risk of bias derived from a lack of blinding to be 'low' because we assumed that the ascertainment of outcomes was not influenced by open (non‐blind) administration. Studies that met the inclusion criteria after screening were included in the review independent of their risk of bias classification. The assigned risk of bias in studies had an influence on the assessment of the certainty of the evidence (GRADE).
Measures of treatment effect
We summarised the effects of dichotomous outcomes (e.g. proportion of patients with hearing improvement measured by pure tone audiometry) as risk ratios (RR) with a 95% confidence interval (CI). For the key dichotomous outcomes presented in the summary of findings tables, we also expressed the results as absolute numbers (the assumed risk in the comparator group and the corresponding risk associated with the experimental intervention, based on its pooled relative effect and 95% CI).
For outcomes measured on a continuous scale (e.g. change in PTA, final PTA), we calculated the mean difference (MD) with a 95% CI. The summary statistic in the meta‐analysis for the primary outcome was the mean difference (MD) of the mean change in dB (baseline/post‐therapy) in hearing threshold between two groups in each study, measured by pure tone audiometry. The summary statistic in the meta‐analysis of the secondary outcome 'final hearing threshold' was the mean difference (MD) of the mean final hearing threshold in dB HL (post‐therapy) between two groups in each study, measured by pure tone audiometry. We used RevMan 5 to compute the measures of treatment effect for each individual study (RevMan 2020).
For hearing outcomes measured on a continuous scale (e.g. change in PTA, final PTA), we assumed a difference of 10 dB to be a clinically relevant effect. This decision was based on the test‐retest reliability of pure tone audiometric measurements, established minimal criteria for improvement in individual patients (Chandrasekhar 2019; Gurgel 2012; Stachler 2012), and on a large RCT on this topic with low risk of bias (Rauch 2011).
For dichotomous outcomes (e.g. proportion of patients with hearing improvement), we used a threshold of 25% or more in RR increase for appreciable benefit as suggested in the GRADE guideline (Guyatt 2011). The 10 dB difference and the 25% criteria were agreed upon by all authors.
Unit of analysis issues
The unit of analysis was the individual participant. We intended to include only studies in which the individual participant was the unit of analysis, regardless of whether they had unilateral or bilateral hearing loss. However, we did identify one study in which the unit of analysis was a single ear (Kosyakov 2011). As only a very small number of participants had bilateral hearing loss, we decided to include these data in the review. Although we were unable to account for the correlation between the ears, treating the data as independent is likely to produce a more conservative estimate of the treatment effect.
Dealing with missing data
We considered missing information about the methods of the included studies (e.g. when the method of randomisation was not reported) in the risk of bias assessment. Where data relating to an outcome of interest were not reported, we contacted the study authors. If the study authors could not provide the missing data or did not respond we excluded the study from the analysis of that outcome. If standard deviation data were not available, we approximated them using standard estimation methods from P values, standard errors or 95% CIs if these were reported, as detailed in the Cochrane Handbook for Systematic Reviews of Interventions (Handbook 2021).
Assessment of heterogeneity
We assessed both clinical and statistical heterogeneity. Clinical heterogeneity may be present even in the absence of statistical heterogeneity. For assessment of clinical heterogeneity we examined the included studies for evidence of major differences in the types of participants recruited, interventions, controls or outcomes measured.
We assessed statistical heterogeneity by visually inspecting the forest plots and by considering the Chi² test and the I² statistic. The latter calculates the percentage of variability that is not due to chance. I² values over 50% suggest the presence of substantial heterogeneity (Handbook 2021). Due to the low power of the Chi² test we set a significance level of P < 0.1.
Assessment of reporting biases
We assessed two aspects of reporting bias: between‐study publication bias and within‐study outcome reporting bias.
Publication bias (between‐study reporting bias)
Where sufficient studies (10 or more) were available for an outcome, we used a funnel plot to assess publication bias.
Outcome reporting bias (within‐study reporting bias)
We assessed within‐study reporting bias by comparing the outcomes reported in the published report with those listed in the methods section. If a study protocol was available, we compared the reported outcomes with the pre‐specified outcomes in the study protocol. When results were not reported in a statistically correct way this was reflected in a designation of high risk of bias due to selective outcome reporting.
Data synthesis
We used RevMan 5 to carry out meta‐analyses (RevMan 2020). Where possible we analysed data to give a summary measure of effect. We always used a fixed‐effect model for meta‐analysis to measure the effect. For dichotomous data, we analysed treatment differences as a risk ratio (RR). For continuous outcomes, if all the data were from the same scale, we pooled mean differences between values obtained at follow‐up and at baseline and reported this as a MD. We performed separate analyses for studies assessing primary and secondary therapy respectively.
Few studies reported the outcomes 'change in hearing threshold with speech audiometry' and 'frequency‐specific hearing loss'. Furthermore, studies often used different methods of speech audiometry and it was not clear if these were directly comparable. Therefore we have not conducted any meta‐analyses for these outcomes, but have instead shown the available data on a forest plot without pooling.
The type of adverse events varied widely between the different studies and it was often unclear whether these events had been systematically assessed and reported. Few studies reported an event rate for each randomised group. More often, a rate was reported for only one group, or a broad statement was made that 'no adverse events were observed'. Sometimes it was unclear to which group a statement applied. The lack of comparable data across groups and across studies meant we were unable to synthesise the data for many types of event and permitted few meta‐analyses.
We considered some adverse events to be directly related to the procedure of intratympanic injection, which may have explained why these events were not always assessed or reported in the comparator group. We considered people with ISSNHL to be very unlikely to experience tympanic membrane perforation, sudden‐onset vertigo (at the time of intratympanic injection) or sudden‐onset ear pain (at the time of intratympanic injection) unless directly attributable to the procedure. For these events we have provided a narrative synthesis of the event rate in the relevant group and presented these as a range.
For adverse events that could feasibly occur in either group, we have only included the studies that provided a rate for both groups, which allowed a comparison between the groups. A full description of all reported adverse event data is available for reference in Table 1, Table 2, Table 3 and Table 4.
Subgroup analysis and investigation of heterogeneity
We did not perform subgroup analysis, due to insufficient data for our planned analyses. We had planned to consider the following subgroups in the review:
-
Degree of hearing loss at initial presentation.
-
Age of patients.
-
Presence of vertigo and/or tinnitus.
-
Time before start of intratympanic treatment.
-
Duration of intratympanic treatment.
-
Drug delivery strategy/system used (e.g. intratympanic injection or continuous delivery etc).
-
Dose of intratympanic treatment.
Sensitivity analysis
We carried out sensitivity analyses to determine whether or not the findings were robust, based on the decisions made in undertaking the review. We planned analyses excluding studies with high risk of bias. Studies with high risk of bias were defined as those that had a high risk of selection bias (bias in randomisation or concealment, or both), an overall loss to follow‐up of > 25%, or unclear or imbalanced baseline parameters (e.g. treatment delay in Ashtiani 2018).
Summary of findings and assessment of the certainty of the evidence
Two independent authors (AL and CM) used the GRADE approach to rate the overall certainty of evidence. The certainty of evidence reflects the extent to which we are confident that an estimate of effect is correct and we considered this in the interpretation of results. There are four possible ratings: high, moderate, low and very low. A rating of high certainty of evidence implies that we are confident in our estimate of effect and that further research is very unlikely to change our confidence in the estimate of effect. A rating of very low certainty implies that any estimate of effect obtained is very uncertain.
The GRADE approach rates evidence from RCTs that do not have serious limitations as high certainty. However, several factors can lead to the downgrading of the evidence to moderate, low or very low. The degree of downgrading is determined by the seriousness of these factors:
-
study limitations (risk of bias);
-
inconsistency (heterogeneity);
-
indirectness of evidence (characteristics of participant population);
-
imprecision (variance of the outcome within studies); and
-
publication bias.
We included a summary of findings table, constructed according to the recommendations described in Chapter 10 of the Cochrane Handbook for Systematic Reviews of Interventions (Handbook 2021), for the following comparisons:
-
Intratympanic corticosteroids versus systemic corticosteroids as primary therapy.
-
Intratympanic plus systemic corticosteroids (combined therapy) versus systemic corticosteroids alone as primary therapy.
-
Intratympanic corticosteroids versus no treatment or versus placebo as secondary therapy.
-
Intratympanic plus systemic corticosteroids (combined therapy) versus systemic corticosteroids alone as secondary therapy.
We included the following outcomes in the summary of findings tables:
-
Change in hearing threshold with pure tone audiometry (pure tone average (PTA)).
-
Proportion of patients whose hearing is improved.
-
Final hearing threshold.
-
Adverse events.
As described above, adverse events were inconsistently reported across the studies, and a wide range of different adverse events were described. For the summary of findings tables, we therefore prioritised events that were considered to be of most relevance to intratympanic injection, namely tympanic membrane perforation, ear pain and vertigo/dizziness. All other adverse events are described in the text of the review and additional tables (Table 1; Table 2; Table 3; Table 4), but not included in the summary of findings tables.
The wording in the comments of the summary of findings tables, in the abstract, the results and the authors’ conclusion sections was based on the 'GRADE guidelines informative statements to communicate the findings of systematic reviews of interventions' (Santesso 2020). In this guideline, producers and users of systematic reviews found statements to communicate findings combining size and certainty of an effect acceptable. The Cochrane Handbook for Systematic Reviews of Interventions (Chapter 15.6.4) also suggests using these narrative statements for drawing conclusions based on the effect estimate from the meta‐analysis and the certainty of the evidence (Handbook 2021).
Results
Description of studies
Results of the search
The flow of records from the number of references identified in the search to the number of studies included in the review is shown in Figure 1. The database search yielded 1720 records after duplicates were removed. We identified 59 additional records through other sources. We screened 1779 records for initial inclusion and discarded 1399 records because they did not study intratympanic corticosteroid treatment of ISSNHL. We discarded a further 328 articles because they were reviews, case reports, study protocols or not randomised controlled studies (i.e. randomised controlled trials and quasi‐randomised controlled trials). Finally, we assessed 52 randomised and quasi‐randomised trials for eligibility. We excluded 20 studies because either the study was carried out in the wrong population (n = 3), or used the wrong intervention (n = 4) or wrong comparator (n = 12). We excluded one study because the duration of follow‐up was seven days or less. See Excluded studies and Characteristics of excluded studies.
Process of selection of studies for inclusion in the review.
Two studies are ongoing. See Characteristics of ongoing studies.
We included the remaining 30 studies in the review.
Included studies
Thirty studies met the criteria for inclusion with 2133 analysed patients in total. See the Characteristics of included studies table for full details.
Study design
All included studies were parallel‐group RCTs. The majority of studies were open‐label trials. Only four studies reported blinding of participants, personnel and outcome assessors (Ashtiani 2018; Battaglia 2008; Plontke 2009; Wu 2011).
Participants
All included studies recruited adult participants. Studies were conducted in a number of locations, including China (Chang 2010; Huang 2021; Li 2011; Peng 2008; Qu 2015; Tong 2021; Wu 2011; Zhou 2011), the Republic of Korea (Ahn 2008; Choi 2011; Hong 2009; Lee 2011; Lim 2013), Greece (Koltsidopoulos 2013; Tsounis 2018; Xenellis 2006), Turkey (Arslan 2011; Ermutlu 2017; Gundogan 2013), Iran (Arastou 2013; Ashtiani 2018), the USA (Battaglia 2008; Rauch 2011), Germany (Plontke 2009), India (Swachia 2016), Italy (Dispenza 2011), Russia (Kosyakov 2011), Saudi Arabia (Al‐Shehri 2015), Sri Lanka (Rupasinghe 2017), and Taiwan (Ho 2004).
Baseline hearing loss
All participants had SSNHL, but the specific hearing threshold required by the studies did differ. The most common threshold was a hearing loss of > 30 dB in three contiguous frequencies, occurring over the course of < 72 hours (Ahn 2008; Choi 2011; Dispenza 2011; Ermutlu 2017; Gundogan 2013; Hong 2009; Huang 2021; Koltsidopoulos 2013; Kosyakov 2011; Lee 2011; Li 2011; Lim 2013; Swachia 2016; Tsounis 2018; Wu 2011; Xenellis 2006; Zhou 2011). Three studies did not describe the use of these thresholds in their methods, but defined SSNHL according to these criteria elsewhere in the article, therefore it is presumed that the same criteria were used (Arastou 2013; Ashtiani 2018; Peng 2008). Some studies used a smaller change in hearing threshold, such as > 20 dB hearing loss in three contiguous frequencies (Arslan 2011; Tong 2021), or > 10 dB hearing loss in three contiguous frequencies (Rupasinghe 2017). Again, one study did not describe the definition of SSNHL in the methods of the paper, but referred elsewhere to a definition of > 20 dB hearing loss in three contiguous frequencies, therefore we presumed this threshold was used (Battaglia 2008).
Two studies required participants to have a pure tone average of 50 dB or higher and the affected ear having hearing at least 30 dB worse than the contralateral (unaffected) ear (Al‐Shehri 2015; Rauch 2011). One study required a hearing threshold of ≥ 50 dB hearing loss for three or more frequencies (PTA including 0.5 kHz, 1 kHz, 2 kHz, 3 kHz and 4 kHz) or ≥ 60 dB for two frequencies, or ≥ 70 dB for any frequency within this range, or a speech reception threshold of ≥ 70 dB SPL or a speech discrimination score of ≥ 30% (Plontke 2009). Three studies did not provide a hearing threshold at which participants were included in the study (Chang 2010; Ho 2004; Qu 2015).
Time to initial treatment
For studies that were concerned with primary treatment, the majority enrolled and commenced treatment within 15 days of the onset of SSNHL (Al‐Shehri 2015; Ashtiani 2018; Choi 2011; Ermutlu 2017; Gundogan 2013; Hong 2009; Huang 2021; Qu 2015; Rauch 2011; Rupasinghe 2017; Swachia 2016; Tong 2021; Tsounis 2018). Five studies permitted enrolment in the study after a longer delay, but most participants were recruited within two weeks (Arslan 2011; Battaglia 2008; Dispenza 2011; Koltsidopoulos 2013; Kosyakov 2011). Four studies did not specify the time from onset of symptoms to treatment as an inclusion criterion. The delay to treatment in three of these studies was a mean of 7 days (Ahn 2008), 8.4 days (Lim 2013), and 5.6 days (Peng 2008). One study specifically recruited participants with poor prognostic factors, which may have included a delay in treatment (Arastou 2013). For this study, 38% of participants had a delay of more than two weeks before receiving their first treatment.
Failure of initial treatment
For studies that were concerned with secondary treatment, participants were recruited based on the failure of initial therapy. Treatment failure was also defined differently across the studies. Two studies based this purely on the improvement in hearing over the course of therapy, with improvement of < 10 dB (Lee 2011) or < 30 dB (Ho 2004) regarded as treatment failure. Three studies considered the difference between ears, with or without the absolute hearing threshold: Wu 2011 used a > 20 dB HL difference on PTA when compared to the contralateral (unaffected) ear to define treatment failure, whilst Li 2011 and Xenellis 2006 both used a > 10 dB HL difference when compared to the contralateral ear or a PTA of < 30 dB. Plontke 2009 required a hearing threshold in the contralateral ear to be at least 20 dB HL better than the affected ear in at least three frequencies between 0.5 kHz and 4 kHz. Zhou 2011 considered treatment failure as a change of less than 15 dB in PTA at four frequencies and an increase of < 15% in speech discrimination score after initial therapy. Chang 2010 stated that participants were included if they were refractory to primary therapy after 20 days, but did not describe how this was defined.
Interventions and comparisons
Most of the included studies investigated primary treatment of ISSNHL (Ahn 2008; Al‐Shehri 2015; Arastou 2013; Arslan 2011; Ashtiani 2018; Battaglia 2008; Choi 2011; Dispenza 2011; Ermutlu 2017; Gundogan 2013; Hong 2009; Huang 2021; Koltsidopoulos 2013; Kosyakov 2011; Lim 2013; Peng 2008; Qu 2015; Rauch 2011; Rupasinghe 2017; Swachia 2016; Tong 2021; Tsounis 2018).
For primary treatment, 12 studies compared intratympanic treatment to systemic steroids, which were predominantly administered orally (Al‐Shehri 2015; Dispenza 2011; Ermutlu 2017; Hong 2009; Huang 2021; Kosyakov 2011; Peng 2008; Qu 2015; Rauch 2011; Rupasinghe 2017; Swachia 2016; Tong 2021). Six studies compared combined treatment (intratympanic plus systemic corticosteroids) with systemic corticosteroids alone (Ahn 2008; Arastou 2013; Arslan 2011; Choi 2011; Gundogan 2013; Koltsidopoulos 2013). Four studies included three treatment arms (intratympanic treatment alone, intratympanic plus systemic treatment and systemic treatment alone) and therefore contributed data to both of these comparisons (Ashtiani 2018; Battaglia 2008; Lim 2013; Tsounis 2018).
A small number of studies investigated secondary treatment, after the failure of initial therapy (Chang 2010; Ho 2004; Lee 2011; Li 2011; Plontke 2009; Wu 2011; Xenellis 2006; Zhou 2011). The type of initial (primary) treatment that participants had received varied, with three studies using intravenous steroids (Plontke 2009; Xenellis 2006; Zhou 2011), three studies using a 10‐ to 14‐day course of oral steroids (Ho 2004; Lee 2011; Li 2011), and one study using an initial dose of intravenous steroids, followed by a tapered oral dose (Wu 2011). One study did not describe the primary therapy that had been used (Chang 2010).
Most studies that considered secondary treatment compared intratympanic steroids to either no treatment (Chang 2010; Ho 2004; Lee 2011; Li 2011; Xenellis 2006), or to placebo (Plontke 2009; Wu 2011). A single study compared intratympanic plus systemic corticosteroids to systemic corticosteroids alone (Zhou 2011).
The nature of the intratympanic injection varied between studies. Most studies used an intratympanic delivery of either dexamethasone or methylprednisolone, administered as a short course of three to four injections, typically over 7 to 14 days. Two studies used notably different methods of administration. One involved daily injections for 10 days, followed by alternate day injections for a further 20 days and ongoing injections twice a week for five months (Kosyakov 2011). One further study used a catheter to provide continuous infusion of dexamethasone over 14 days, rather than intermittent injections (Plontke 2009). In one study the duration of systemic treatment (15 days) was much shorter than that of the intratympanic treatment (six months) (Kosyakov 2011).
Outcomes
Duration of follow‐up varied across the studies. Four studies followed participants for 15 days or less (Arastou 2013; Arslan 2011; Plontke 2009; Qu 2015). Six studies followed participants for between 16 and 30 days (Ashtiani 2018; Chang 2010; Gundogan 2013; Lim 2013; Peng 2008; Wu 2011). The remaining studies followed participants for over one month, with a range of 38 days (Tong 2021) to 204 days (Dispenza 2011). Huang 2021 followed participants for 90 days, but we extracted data for change in hearing thresholds from an interim analysis at 12 days because the comparison of interest (intratympanic versus systemic corticosteroid) was administered only up to that point in time. All other outcome data were reported at the longest follow‐up point for each study, except instances where there was very high dropout at the final time point (as described in Included studies).
Change in hearing threshold with pure tone audiometry (pure tone average)
Most studies assessed hearing thresholds with a pure tone average based on four frequencies, either 0.5 kHz, 1 kHz, 2 kHz or 4 kHz (Al‐Shehri 2015; Arslan 2011; Choi 2011; Dispenza 2011; Kosyakov 2011; Li 2011; Peng 2008; Rauch 2011; Swachia 2016; Tsounis 2018; Wu 2011; Xenellis 2006; Zhou 2011), or 0.5 kHz, 1 kHz, 2 kHz and 3 kHz (Ahn 2008; Gundogan 2013; Hong 2009; Lee 2011; Lim 2013; Plontke 2009). Two studies used the average of three frequencies (0.5 kHz, 1 kHz, 2 kHz, Battaglia 2008; Ermutlu 2017), two studies used five frequencies (0.25 kHz, 0.5 kHz, 1 kHz, 2 kHz and 4kHz) (Arastou 2013; Ashtiani 2018), and four studies used six or more frequencies (Chang 2010; Ho 2004; Koltsidopoulos 2013; Tong 2021). Two studies did not describe the frequencies used (Qu 2015; Rupasinghe 2017).
Proportion of patients whose hearing is improved
We included data for this outcome regardless of the definition of 'improvement' used in the individual studies. However, this definition was not consistent across the different studies. A number of studies used the criterion of a specific change in hearing threshold to identify those who had improved. This was typically a change of at least 10 dB (Arslan 2011; Dispenza 2011; Ho 2004; Lee 2011; Li 2011; Lim 2013; Rauch 2011; Swachia 2016; Tong 2021; Wu 2011; Xenellis 2006), or 15 dB (Arastou 2013; Kosyakov 2011; Peng 2008; Qu 2015; Tsounis 2018; Zhou 2011), over the follow‐up period. Four studies used Siegel's criteria to assess improvement (Siegel 1975), where 'any' improvement is considered to be > 15 dB change in hearing threshold and final hearing threshold ≥ 75 dB (Ahn 2008; Choi 2011; Gundogan 2013; Hong 2009). Four studies considered both the change in hearing threshold and improvement in word recognition scores (WRS) or speech discrimination scores (SDS) when assessing improvement: Ashtiani 2018 (> 10 dB decrease in PTA or > 15% improvement in SDS), Battaglia 2008 (> 15 dB decrease in PTA or > 25% improvement in SDS), Ermutlu 2017 (> 10 dB decrease in PTA or > 10% improvement in WRS) and Koltsidopoulos 2013 (> 10 dB decrease in PTA and 15% improvement in SDS). One study used a decrease of > 30 dB in PTA, or an assessment of recovery to 50% of maximum possible (as compared to the unaffected ear) to indicate improvement (Plontke 2009). One study reported on improvement in hearing, but did not describe the criteria (Rupasinghe 2017). Two studies did not assess improvement (Al‐Shehri 2015; Chang 2010).
Final hearing threshold with pure tone audiometry
Frequencies used to assess this outcome were identical to those used for change in hearing threshold (see above).
Change in hearing threshold with speech audiometry
This outcome was only reported by a small number of studies, and was assessed with a variety of instruments, including speech discrimination scores, speech reception thresholds and word recognition scores. Considering the different metrics used to measure this outcome, and a concern that assessment conducted in different languages may not be directly comparable, we did not conduct any meta‐analyses. Six studies reported changes from baseline or final values for speech discrimination score (Ashtiani 2018; Battaglia 2008; Gundogan 2013; Koltsidopoulos 2013; Plontke 2009; Zhou 2011). Two studies reported changes from baseline in speech reception threshold (Ashtiani 2018; Plontke 2009), and one study reported change from baseline in the word recognition score (Rauch 2011).
Frequency‐specific changes with pure tone audiometry
Again, few studies reported on frequency‐specific changes with pure tone audiometry. There was also inconsistency in the frequencies that were assessed, and some studies presented pooled data across a small number of frequencies (low, mid and high), rather than reporting individual frequencies. Therefore we did not conduct any meta‐analyses for this outcome. The only studies assessing this were: Ahn 2008; Arslan 2011; Dispenza 2011; Gundogan 2013; Hong 2009; Huang 2021; Kosyakov 2011 Lee 2011; Lim 2013 and Tong 2021.
Mean level of improvement in those whose hearing is improved
This outcome was not assessed or reported by any of the included studies.
Percentage of patients reaching serviceable hearing (for those with profound pre‐treatment hearing loss)
This outcome was not assessed or reported by any of the included studies.
Effect on tinnitus and vertigo
This outcome was not assessed or reported by any of the included studies. Some studies reported on tinnitus and vertigo, but as adverse effects of the intervention, rather than assessing whether the intervention may have a beneficial effect on existing symptoms.
Minor and serious adverse effects
As described in Data synthesis, the adverse effects reported by the individual studies were wide‐ranging. The only adverse effects that were consistently reported across a large number of studies were those directly related to intratympanic injection (including persistent tympanic membrane perforation, pain or dizziness/vertigo at the time of the injection). However, as these events were clearly related to the intratympanic injection, and would not occur if participants received no treatment or systemic steroids, we considered it inappropriate to report a risk ratio comparing the intervention and comparator groups. Instead we have reported the rate of these complications for those individuals who received intratympanic injections. Specific details on other adverse events are included in the Effects of interventions and Table 1; Table 2; Table 3 and Table 4.
Excluded studies
See Characteristics of excluded studies.
Five randomised controlled trials compared the efficacy of intratympanic corticosteroid therapy in combination with hyperbaric oxygen therapy (HBO) (Attanasio 2015; Cho 2018; Gui‐li 2018; Sevil 2016; Zhou 2006), and one study used HBO treatment as a comparator (Cvorovic 2013). Since data suggest that HBO itself might have an effect on hearing recovery (Bennett 2012), and since the addition of HBO was not part of the interventions to be studied in this review (see methods), we excluded these studies.
Amizadeh 2021 compared combined corticosteroid treatment and systemic corticosteroid treatment as primary intervention. The study was excluded because the route of administration and dosage of systemic corticosteroid differed between groups.
The study Rogha 2017 and the trial registration NCT04766853 compared corticosteroid treatment by intratympanic injection of dexamethasone with intratympanic injection of dexamethasone mixed with hyaluronic acid. This type of comparison was not part of the review.
Chang 2020 compared intratympanic corticosteroid treatment with ear drop corticosteroid treatment as a primary intervention. This type of comparison was not part of the review.
Han 2021 compared intratympanic injection of corticosteroid versus corticosteroid administered via endoscopic tympanoplasty. This type of comparison was not part of the review.
We excluded the randomised controlled studies Berjis 2016 and Sun 2016 because two intratympanic treatment protocols using two different intratympanically applied corticosteroids were compared. This type of comparison was not part of the review.
We excluded Li 2016 because it compared intratympanic corticosteroid treatment with intratympanic corticosteroid plus mouse nerve growth factor treatment. This type of comparison was not part of the review.
Song 2018 compared intratympanic corticosteroid treatment with postauricular injection of corticosteroid as primary Intervention. This type of comparison was not part of the review.
In the study Park 2011, two methods of combination (intratympanic and systemic) therapy were compared. In the simultaneous intratympanic dexamethasone group, local drug application was given initially (as primary therapy for ISSNHL) with systemic steroids (intravenous dexamethasone followed by oral prednisolone). In the other "subsequent intratympanic dexamethasone group", intratympanic dexamethasone was given seven days after systemic treatment. There was no control group for the intratympanic salvage situation without local application. This type of comparison was not part of the review.
Filipo 2013 compared intratympanic corticosteroid treatment with intratympanic placebo as a primary intervention. The study endpoint was seven days after the start of treatment. Studies with a study endpoint of seven days or less after start of treatment were excluded from the review.
Choo 2017 compared intratympanic corticosteroid treatment, oral corticosteroid treatment and combined corticosteroid treatment as a primary intervention separated by low‐ or high‐frequency hearing loss. A comparison of hearing improvement in ISSNHL patients with low‐ and high‐frequency hearing loss was not part of the review.
We excluded Chen 2015 because the comparison group in this study included a mixture of patients receiving treatments with systemic steroids or systemic steroids plus intratympanic steroids.
We excluded Diao 2012 because the study population included a high proportion of patients with bilateral sudden hearing loss that raises doubt about whether they represented people with ISSNHL. Further, the unit of analysis in this study was ears instead of participants, as is used in this review.
Ongoing studies
Wang 2021 is a non‐blinded, parallel‐group randomised controlled trial that is being carried out in China, from October 2020. It compares nine intratympanic injections of dexamethasone over 14 days with daily oral prednisolone over 14 days for the primary treatment of ISSNHL. Pure tone thresholds, speech recognition, vestibular evoked myogenic potentials (VEMPs), Dizziness Handicap Inventory (DHI) and Tinnitus Handicap Inventory (THI) will be measured before treatment and one month after termination of treatment.
The study Yang 2020 is a non‐blinded, parallel‐group, randomised superiority trial that is being carried out in China, from January 2018. It compares four intratympanic injections of methylprednisolone over one week with daily intravenous methylprednisolone over five days for the primary treatment of ISSNHL in patients with diabetes mellitus. Pure tone thresholds will be measured before treatment and one month after termination of treatment. Secondary outcome measures will include the pure tone average at three months after treatment and blood glucose changes during treatment.
Risk of bias in included studies
We deemed the risk of bias to be generally rather high in most of the included studies. We assigned only four of the 30 included studies an overall low risk of bias (Plontke 2009; Rauch 2011; Tsounis 2018; Wu 2011). An overview of the risk of bias for each included study is provided in Figure 2. Figure 3 presents the proportion of each risk of bias domain that we found to be high risk/low risk/unclear risk across the whole review.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Allocation
Randomisation was adequate in nine studies (Arastou 2013; Ashtiani 2018; Gundogan 2013; Kosyakov 2011; Plontke 2009; Qu 2015; Tsounis 2018; Rauch 2011; Wu 2011). Ahn 2008 also used computerised random allocation, but patients who refused the allocated therapy were excluded. Twelve studies stated that they were randomised without details concerning the methods of randomisation (Al‐Shehri 2015; Battaglia 2008; Chang 2010; Choi 2011; Dispenza 2011; Ermutlu 2017; Ho 2004; Hong 2009; Lee 2011; Li 2011; Swachia 2016; Xenellis 2006). Eight studies used inadequate randomisation methods (randomisation according to sequence of admission or actual date) (Arslan 2011; Huang 2021; Koltsidopoulos 2013; Lim 2013; Peng 2008; Rupasinghe 2017; Tong 2021; Zhou 2011).
Allocation concealment was adequate in only five studies (Ashtiani 2018; Plontke 2009; Rauch 2011; Tsounis 2018; Wu 2011). The other studies reported either an inadequate method of concealment or it was not mentioned at all. In studies that used sequence of admission or actual date for randomisation adequate allocation concealment is not possible (Arslan 2011; Huang 2021; Koltsidopoulos 2013; Lim 2013; Peng 2008; Rupasinghe 2017; Tong 2021; Zhou 2011).
Blinding
Only four studies used placebo therapy with blinding of participants and personnel during their trials (Ashtiani 2018; Battaglia 2008; Plontke 2009; Wu 2011). Twenty‐six studies were not placebo‐controlled. In one, it was explicitly stated that participants and personnel were not blinded (Tsounis 2018); in the others we assumed that neither were blinded. We judged the risk of bias in non‐placebo‐controlled studies to be generally 'low' assuming that the outcomes were not influenced by open administration of study therapy. In two studies patients in the intratympanic treatment group could refuse the therapy after allocation, so we deemed the risk of bias to be high (Ahn 2008; Zhou 2011).
In 13 studies, we considered the risk of detection bias to be low because they were either placebo‐controlled, or there was blinding of outcome assessment (Al‐Shehri 2015; Arastou 2013; Ashtiani 2018; Battaglia 2008; Hong 2009; Koltsidopoulos 2013; Kosyakov 2011; Lim 2013; Rauch 2011; Plontke 2009; Tong 2021; Tsounis 2018; Wu 2011). Seventeen studies gave no information on blinding of outcome assessment and the risk of bias was unclear.
Incomplete outcome data
Fourteen studies reported results for all randomised participants. The study Rauch 2011 reported a low dropout rate below 5%. We classified six studies as high risk of bias because the dropout rate was higher than 10% (Ashtiani 2018; Battaglia 2008; Ermutlu 2017; Hong 2009; Rupasinghe 2017; Tsounis 2018), and we classified three studies as high risk because dropouts were not balanced across treatment arms (Dispenza 2011) or were related to the therapy (Wu 2011; Zhou 2011). The studies Huang 2021, Gundogan 2013, Plontke 2009 and Tong 2021 reported a moderate rate of dropout (5% to 10%) but it was balanced across treatment arms. Ahn 2008 and Chang 2010 gave insufficient information to permit judgement as they did not report the number of participants that were analysed.
Selective reporting
Twenty‐one studies specified and reported the main outcome measures (Ahn 2008; Al‐Shehri 2015; Arslan 2011; Ashtiani 2018; Chang 2010; Ermutlu 2017; Gundogan 2013; Hong 2009; Huang 2021; Kosyakov 2011; Li 2011; Lim 2013; Peng 2008; Plontke 2009; Qu 2015; Rauch 2011; Swachia 2016; Tong 2021; Tsounis 2018; Wu 2011; Xenellis 2006). There was no indication of selective reporting in these studies. Four studies failed to report the standard deviation for change in hearing threshold (Arastou 2013; Battaglia 2008; Choi 2011; Lee 2011). Some studies included several follow‐up time points but results were only shown for the last time point (Choi 2011; Dispenza 2011; Ho 2004; Lee 2011; Zhou 2011). In two studies contradictions were present between pre‐specified outcome parameters in the methods section and the presented outcomes in the results: in Koltsidopoulos 2013 a 7PTA was defined as the primary outcome parameter but the reported hearing loss before treatment and hearing improvement were both based on a 4PTA. In the study Tong 2021, a decrease in PTA of > 30 dB HL was a criterion for successful treatment, but a > 10 dB HL decrease was actually used. In Rupasinghe 2017, the criteria for hearing improvement were not reported.
Other potential sources of bias
A common source of bias was imbalance between groups for a number of factors: there were often unexplained differences in the number of participants in each group (Arastou 2013; Arslan 2011; Choi 2011; Lee 2011; Li 2011; Qu 2015; Swachia 2016), and differences between groups in the delay before commencing treatment (Battaglia 2008; Dispenza 2011; Lim 2013; Rupasinghe 2017; Xenellis 2006). In two studies there was a difference between groups in the length of treatment (Kosyakov 2011; Peng 2008). In one study, follow‐up was longer in the intervention group than in the comparator group (Lim 2013). Only the studies Ashtiani 2018, Koltsidopoulos 2013, Lim 2013, Plontke 2009 and Tsounis 2018 performed a sample size determination. Other studies either did not do so (Ahn 2008; Al‐Shehri 2015; Arastou 2013; Arslan 2011; Chang 2010; Choi 2011; Dispenza 2011; Ermutlu 2017; Gundogan 2013; Ho 2004; Hong 2009; Huang 2021; Kosyakov 2011; Lee 2011; Li 2011; Peng 2008; Qu 2015; Rupasinghe 2017; Swachia 2016; Tong 2021; Wu 2011; Xenellis 2006; Zhou 2011), or terminated recruitment before reaching a sufficient number of participants based on sample size calculation (Battaglia 2008). In consequence, the number of included participants per treatment arm was small in most studies. Studies without sample size determination are prone to type II errors. This is not discussed in any of these studies. Small study populations are also prone to imbalances between treatment arms in terms of potential confounding factors, including the propensity in some patients with ISSNHL towards spontaneous hearing recovery.
A broad range of delay between the onset of symptoms and the start of treatment was evident in some studies (Arastou 2013; Arslan 2011; Battaglia 2008; Peng 2008; Xenellis 2006). Treatment delay is recognised as one of the main factors that influences the observed hearing improvement (Liebau 2017). When there is a small number of participants per group, studies may differ in this respect across treatment arms. As noted above, a noticeable difference in treatment delay was evident in a number of studies (Battaglia 2008; Dispenza 2011; Lim 2013; Rupasinghe 2017; Xenellis 2006). This could have influenced the reported difference in outcome between treatment arms.
The main baseline parameter that influences the observed hearing improvement is the level of hearing loss of the patient at the beginning of the observation period (Liebau 2017). Although a balanced hearing loss before treatment between intervention arms is extremely important, in two studies hearing loss before treatment differed by more than 10 dB HL across groups (Choi 2011; Dispenza 2011). The differences in hearing loss before treatment between the intervention arms may have influenced the reported difference in outcome.
Many studies did not report the baseline characteristics of their treatment arms, or reported them inadequately. In 13 studies the treatment delay in each treatment arm was not reported (Al‐Shehri 2015; Ashtiani 2018; Chang 2010; Ho 2004; Kosyakov 2011; Lee 2011; Li 2011; Qu 2015; Rupasinghe 2017; Swachia 2016; Wu 2011; Xenellis 2006; Zhou 2011), or a standard deviation is missing in that parameter (Dispenza 2011; Hong 2009). In the study Rupasinghe 2017, the hearing thresholds before treatment in participants per group are not reported and Dispenza 2011 omitted the standard deviation for that parameter. The studies Zhou 2011 and Ho 2004 reported the mean hearing loss before treatment in the intervention group but not in the control group.
The studies Arslan 2011, Plontke 2009, Qu 2015 and Zhou 2011 conducted a very short follow‐up (two weeks or less), which could result in a bias in the estimation of treatment effects. In Huang 2021, the comparison of interest for this review (intratympanic corticosteroid versus systemic corticosteroid) was observed for only 12 days from the start of treatment, the comparator group receiving systemic treatment then switching to intratympanic corticosteroid. As noted above, in Lim 2013 the duration of follow‐up was longer in the intervention group in comparison to the control group (21 versus 17 days). This could underestimate the treatment effect in the control group. In Peng 2008, the duration of treatment differed between treatment arms (17 versus 27 days). This was also true of Kosyakov 2011 (six months versus 15 days). In each case, the discrepancy in the duration of treatment could bias the estimate of effect. Rupasinghe 2017 included patients with very mild hearing loss (> 10 dB HL) and Kosyakov 2011 included only mild cases of ISSNHL. By contrast, Zhou 2011 included only patients with poor prognosis (see the inclusion criteria of the study for details). All three studies risk bias due to selection of the study population. In Xenellis 2006, the omission of overall hearing improvement in the control group is unlike the other studies but is not discussed by the authors.
Effects of interventions
See: Summary of findings 1 Intratympanic corticosteroids versus systemic corticosteroids as primary therapy; Summary of findings 2 Intratympanic plus systemic corticosteroids (combined therapy) versus systemic corticosteroids alone as primary therapy; Summary of findings 3 Intratympanic corticosteroids versus no treatment or versus placebo as secondary therapy; Summary of findings 4 Intratympanic plus systemic corticosteroids (combined therapy) versus systemic corticosteroids alone as secondary therapy
Intratympanic corticosteroids versus no treatment or versus placebo as primary therapy
No study compared the effects of intratympanic corticosteroids versus no treatment or placebo on hearing improvement for primary therapy of ISSNHL.
Intratympanic corticosteroids versus systemic corticosteroids as primary therapy
Sixteen studies compared the efficacy of a primary intratympanic corticosteroid treatment with systemic corticosteroid treatment (Al‐Shehri 2015; Ashtiani 2018; Battaglia 2008; Dispenza 2011; Ermutlu 2017; Hong 2009; Huang 2021; Kosyakov 2011; Lim 2013; Peng 2008; Qu 2015; Rauch 2011; Rupasinghe 2017; Swachia 2016; Tong 2021; Tsounis 2018).
Change in hearing threshold with pure tone audiometry (pure tone average)
Ten studies (701 participants) were included in the meta‐analysis. Battaglia 2008 could not be included as no variance was reported. The mean change in PTA between baseline and 17 to 182 days (range) after start of therapy in participants with intratympanic therapy was ‐5.93 dB (95% confidence interval (CI) ‐7.61 to ‐4.26; 701 participants; 10 studies; I2 = 80%; low‐certainty evidence) (Analysis 1.1). The point estimate of effect did not exceed the minimally important difference of ‐10 dB, however. Primary intratympanic therapy may result, therefore, in little to no improvement in hearing threshold compared to systemic steroids.
Although we noted high heterogeneity, we considered this unlikely to affect the conclusion of the analysis, as most studies resulted in an estimated effect size that was of borderline clinical significance (did not exceed the minimally important difference). Rauch 2011 and Tsounis 2018 were included in the pre‐planned sensitivity analysis. These two studies (319 participants) found that the mean change of PTA in participants with intratympanic treatment was lower compared with the systemic treatment group but did not exceed the minimally important difference of 10 dB (mean difference (MD) 2.00, 95% CI ‐2.79 to 6.79; 319 participants; 2 studies; I2 = 0%). The sensitivity analysis therefore confirmed the result from our primary analysis. As the treatment regime for Kosyakov 2011 was extremely different from all other studies (six months of intratympanic corticosteroid), and as the duration of treatment differed so markedly between treatment arms (the comparator group receiving only 15 days of systemic corticosteroid), we conducted an additional sensitivity analysis to investigate the degree to which excluding this study would impact the pooled estimate. After its exclusion, the pooled mean difference again only indicated a trivial effect in favour of intratympanic therapy (MD ‐2.81, 95% CI ‐4.49 to ‐0.66; 651 participants; 9 studies; I2 =67%; low‐certainty evidence). We concluded that the inclusion of this study in the meta‐analysis did not affect the pooled effect estimate to any serious degree.
Proportion of patients whose hearing is improved
Fourteen studies (972 participants) were included in the meta‐analysis. In one study there was high (35.1%) loss to follow‐up at the final (three‐month) assessment (Rupasinghe 2017). We therefore extracted outcome data from the penultimate assessment at one month. A slightly higher proportion of participants with intratympanic therapy had improved hearing at 10 to 182 days (range) after the start of treatment compared with the systemic treatment group. The point estimate of effect did not, however, exceed the minimally important difference of 25% and the 95% confidence interval included no difference between groups (risk ratio (RR) 1.04, 95% CI 0.97 to 1.12; 972 participants; 14 studies; I2 = 16%) (Analysis 1.2). Primary intratympanic therapy, therefore, probably results in little to no difference in the proportion of patients whose hearing is improved compared to systemic corticosteroids.
The studies Ashtiani 2018, Rauch 2011 and Tsounis 2018 were included in the sensitivity analysis. The three studies (396 participants) found that fewer participants with intratympanic treatment had improvement of hearing compared with the systemic treatment group but the difference did not exceed the minimally important difference of 25% (RR 0.96, 95% CI 0.86 to 1.07; 396 participants; 3 studies; I2 = 16%). The sensitivity analysis therefore confirmed the result from our primary analysis.
Final hearing threshold with pure tone audiometry (pure tone average)
Seven studies (516 participants) were included in the meta‐analysis. The study Tsounis 2018 could not be included as no variance was reported. The final PTA at 17 to 183 days (range) after the start of therapy in participants with intratympanic therapy was lower (better) compared with the systemic treatment group. The point estimate of effect did not, however, exceed the minimally important difference of 10 dB HL (MD ‐3.31, 95% CI ‐6.16 to ‐0.47; 516 participants; 7 studies; I2 = 41%; low‐certainty evidence) (Analysis 1.3). Primary intratympanic therapy may result, therefore, in little to no difference in the final hearing threshold compared to systemic corticosteroids. The study Rauch 2011 was included in the sensitivity analysis and confirmed this result (MD 1.60, 95% CI ‐5.75 to 8.95; 250 participants; 1 study).
Change in hearing threshold with speech audiometry
Only one study reported on the change in hearing threshold with speech audiometry (Ashtiani 2018). The speech reception threshold may be lower (better) in the group who received intratympanic corticosteroids compared to those who received systemic corticosteroids, although the confidence interval crosses unity (MD ‐8.85 dB, 95% CI ‐19.58 to 1.88; 98 participants; 1 study) (Analysis 1.4).
Three further studies used methods other than hearing threshold to assess speech audiometry (Ashtiani 2018; Battaglia 2008; Rauch 2011). These studies reported on the change in speech recognition or discrimination, using either a speech discrimination score (SDS) or a word recognition score (WRS). Although these did not relate to our pre‐specified outcome (change in hearing threshold with speech audiometry) we considered that they were assessing the same underlying outcome (speech audiometry), therefore we have included them for completeness. Due to the different assessment tools used, inconsistency in terminology (making it unclear whether measures were comparable), and the different languages in which these were conducted, we considered that it was not appropriate to pool the data. The results of these additional measures are presented in Analysis 1.5.
The study Ermutlu 2017 performed speech audiometry (speech recognition threshold, word recognition score) but did not report results separately for these tests ‐ they were instead reported as part of a composite outcome for 'recovery'. Therefore this study was not included for this outcome.
Frequency‐specific changes with pure tone audiometry
Five studies reported on frequency‐specific changes with pure tone audiometry (Dispenza 2011; Hong 2009; Huang 2021; Lim 2013; Tong 2021). Due to the different frequencies assessed in the studies, and heterogeneity in the effect estimates where studies did assess the same frequency, we did not meta‐analyse these data. Instead the results from each study are shown in Analysis 1.6.
Mean level of improvement, in those whose hearing is improved
This outcome was not reported by any of the included studies.
Percentage of patients reaching serviceable hearing
This outcome was not reported by any of the included studies.
Effect on tinnitus and vertigo
This outcome was not reported by any of the included studies.
Minor and serious adverse events
For this comparison, 12 studies provided information about adverse events (Al‐Shehri 2015; Dispenza 2011; Ermutlu 2017; Hong 2009; Huang 2021; Kosyakov 2011; Qu 2015; Rauch 2011; Rupasinghe 2017; Swachia 2016; Tong 2021; Tsounis 2018). In four studies, reporting was incomplete, either because a rate was not provided for both randomised groups (Kosyakov 2011; Swachia 2016), or because it was unclear in which group (or groups) events were observed (Qu 2015; Tsounis 2018). In one additional study adverse event data were not reported specifically for the period of follow‐up during which the allocated interventions matched the comparison of interest for this review (Huang 2021). Despite these limitations in the reporting of adverse events, meta‐analysis was possible for some adverse event outcomes (Analysis 1.7).
Persistent tympanic membrane perforation
Four studies reported a rate of tympanic membrane perforation of between 0% (0/30) and 3.9% (5/129) for those who received an intratympanic corticosteroid injection (Huang 2021; Kosyakov 2011; Rauch 2011; Tong 2021). Note that in one study both groups received intratympanic injection: one group received intratympanic corticosteroid, and the other received intravenous followed by intratympanic corticosteroid (Huang 2021). We concluded that the evidence is very uncertain regarding the risk of tympanic membrane perforation for those who received intratympanic corticosteroid as primary treatment (463 participants; 4 studies; very low‐certainty evidence).
Vertigo/dizziness, timing not reported
A single study provided a comparison between intratympanic and systemic corticosteroid, resulting in a risk ratio of 2.53 (95% CI 1.41 to 4.54) (Rauch 2011). It is not specified whether all of the patients in the intratympanic corticosteroid group experiencing vertigo did so at the time of injection. We concluded that intratympanic therapy may increase the risk of vertigo/dizziness of unspecified timing as compared to systemic corticosteroid (250 participants; 1 study; low‐certainty evidence).
Vertigo/dizziness at the time of intratympanic injection
Four studies reported a rate of vertigo/dizziness of between 1.5% (1/67) and 21% (4/19) for those who received an intratympanic injection (Ermutlu 2017; Huang 2021; Tong 2021; Tsounis 2018). We have included in this analysis all participants who received an intratympanic injection in these studies. For two studies, this included participants in another treatment arm. In Huang 2021, participants in the control arm also received intratympanic injection at a later point in the trial. In Tsounis 2018, data were reported for participants who received intratympanic corticosteroids alone, and combined therapy with intratympanic corticosteroids and systemic corticosteroids. We concluded that the evidence is very uncertain regarding the risk of vertigo/dizziness at the time of intratympanic corticosteroid treatment as primary therapy (301 participants; 4 studies; very low‐certainty evidence).
Ear pain, timing not reported
Two studies contributed data to a meta‐analysis (Al‐Shehri 2015; Rauch 2011). The risk ratio was 15.68 (95% CI 6.22 to 39.49), favouring systemic corticosteroid. In each study, the number of participants with ear pain/earache was presented separately from the numbers with ear pain at intratympanic injection. It was assumed, therefore, that those participants with pain at injection were not included among those with ear pain/earache. We concluded that intratympanic corticosteroid injection probably increases the risk of ear pain as compared to systemic corticosteroid when used as primary treatment (289 participants; 2 studies; moderate‐certainty evidence).
Ear pain at the time of injection
Three studies reported a rate of ear pain from 4.8% (5/104) to 27.1% (35/129) (Al‐Shehri 2015; Huang 2021; Rauch 2011). In Al‐Shehri 2015 and Rauch 2011, the number of participants with ear pain/earache was presented separately from the numbers with ear pain at intratympanic injection. It was assumed, therefore, that those participants with pain at injection were not included among those with ear pain/earache. The evidence suggests that there is a risk of ear pain at the time of intratympanic injection of corticosteroid as primary treatment (393 participants; 3 studies; low‐certainty evidence).
Mood change
Two studies contributed data to a meta‐analysis (Al‐Shehri 2015; Rauch 2011). The risk ratio was 0.22 (95% CI 0.13 to 0.37), favouring intratympanic corticosteroid. We concluded that intratympanic corticosteroid likely results in a large reduction in risk compared to systemic corticosteroids (289 participants; 2 studies; moderate‐certainty evidence).
Blood glucose problems
Two studies contributed data to a meta‐analysis (Al‐Shehri 2015; Rauch 2011). The risk ratio was 0.54 (95% CI 0.35 to 0.85), favouring intratympanic corticosteroid. We concluded that intratympanic corticosteroid may result in a reduction in risk compared to systemic corticosteroid (289 participants; 2 studies; low‐certainty evidence).
Sleep change
Two studies contributed data to a meta‐analysis (Al‐Shehri 2015; Rauch 2011). The risk ratio was 0.19 (95% CI 0.10 to 0.36), favouring intratympanic corticosteroid. We concluded that intratympanic corticosteroid likely results in a large reduction in risk compared to systemic corticosteroid (289 participants; 2 studies; moderate‐certainty evidence).
Appetite change
Two studies contributed data to a meta‐analysis (Al‐Shehri 2015; Rauch 2011). The risk ratio was 0.20 (95% CI 0.09 to 0.44), favouring intratympanic corticosteroid. We concluded that intratympanic corticosteroid likely results in a large reduction in risk compared to systemic corticosteroid (289 participants; 2 studies; moderate‐certainty evidence).
Weight change
Two studies contributed data to a meta‐analysis (Al‐Shehri 2015; Rauch 2011). The risk ratio was 0.28 (95% CI 0.13 to 0.61), favouring intratympanic corticosteroid. We concluded that intratympanic corticosteroid likely results in a large reduction in risk compared to systemic corticosteroid (289 participants; 2 studies; moderate‐certainty evidence).
Dry mouth
Two studies contributed data to a meta‐analysis (Al‐Shehri 2015; Rauch 2011). The risk ratio was 0.15 (95% CI 0.06 to 0.35), favouring intratympanic corticosteroid. We concluded that intratympanic corticosteroid likely results in a large reduction in risk compared with systemic corticosteroids (289 participants; 2 studies; moderate‐certainty evidence).
Otitis media
One study reported a rate for each group (Rauch 2011). The risk ratio was 3.28 (95% CI 0.70 to 15.49), favouring systemic corticosteroid. We concluded that intratympanic corticosteroid may result in a large increase in risk compared to systemic corticosteroid (250 participants; 1 study; low‐certainty evidence).
Table 1 provides details of the more limited data on other reported adverse events.
Intratympanic plus systemic corticosteroids (combined therapy) versus no treatment or versus placebo as primary therapy
No study compared the effects of intratympanic corticosteroids plus systemic corticosteroids (combined therapy) versus no treatment or versus placebo on hearing improvement for primary therapy of ISSNHL.
Intratympanic plus systemic corticosteroids (combined therapy) versus systemic corticosteroids alone as primary therapy
Ten studies compared the efficacy of a primary combined therapy with a systemic corticosteroid therapy (Ahn 2008; Arastou 2013; Arslan 2011; Ashtiani 2018; Battaglia 2008; Choi 2011; Gundogan 2013; Koltsidopoulos 2013; Lim 2013; Tsounis 2018). Three of our secondary outcome measures were not reported by any of the included studies.
Change in hearing threshold with pure tone audiometry (pure tone average)
Six studies (435 participants) were included in the meta‐analysis. The study Battaglia 2008 could not be included as no variance was reported. The mean change in PTA between baseline and 15 to 91 days (range) after start of therapy improved more in participants with combined therapy. The point estimate of effect did not exceed the minimally important difference of ‐10 dB (MD ‐8.55, 95% CI ‐12.48 to ‐4.61; 435 participants; 6 studies; I2 = 32%; low‐certainty evidence) (Analysis 2.1). Primary combined therapy may result, therefore, in a slight improvement in hearing threshold compared to systemic corticosteroids alone, but it is not certain whether the extent of improvement would be meaningful to people with ISSNHL. The study Tsounis 2018 was included in the sensitivity analysis and confirmed this result (MD 0.80, 95% CI ‐8.41 to 10.01; 68 participants; 1 study).
Proportion of patients whose hearing is improved
Ten studies (788 participants) were included in the meta‐analysis. A higher proportion of participants with combined therapy had improvement of hearing at 15 to 91 days (range) after start of treatment. The point estimate of effect exceeded the minimally important difference of 25% (RR 1.27, 95% CI 1.15 to 1.41; 788 participants; 10 studies; I2 = 47%) (Analysis 2.2). Primary combined therapy may, therefore, increase the proportion of patients whose hearing is improved compared to systemic corticosteroids alone (low‐certainty evidence). The studies Ashtiani 2018 and Tsounis 2018 were included in the sensitivity analysis and found that more participants with combined therapy had improvement of hearing but this did not exceed the minimally important difference of 25% (RR 1.08, 95% CI 0.88 to 1.33; 148 participants; 2 studies; I2 = 0%).
Final hearing threshold with pure tone audiometry (pure tone average)
Three studies (194 participants) were included in the meta‐analysis. The studies Arastou 2013 and Tsounis 2018 could not be included since no variance was reported. The final PTA at 15 to 56 days (range) after start of therapy in participants with combined therapy was lower (more favourable) when compared with the systemic treatment group but the point estimate of effect did not exceed the minimally important difference of ‐10 dB (MD ‐9.11, 95% CI ‐1.67 to ‐16.56; 194 participants; 3 studies; I2 = 35%; very low‐certainty evidence) (Analysis 2.3). Primary combined therapy may result, therefore, in slightly lower (better) final hearing thresholds compared to systemic corticosteroids alone, but the evidence is very uncertain. No eligible studies could be identified for a sensitivity analysis.
Change in hearing threshold with speech audiometry
One study reported on the change in speech recognition threshold (Ashtiani 2018). The speech reception threshold may be lower (better) in the group who received combination treatment compared to those who received systemic corticosteroids, although the confidence interval crosses unity (mean difference ‐7.59 dB; 95% CI ‐20.22 to 5.04; 98 participants; 1 study) (Analysis 2.4).
Four studies reported on alternative measures of speech audiometry (Ashtiani 2018; Battaglia 2008; Gundogan 2013; Koltsidopoulos 2013). The results of these additional measures are presented in Analysis 2.5. All had an effect direction that favoured combined treatment. Koltsidopoulos 2013 reported on the change in speech discrimination score using medians and an interquartile range, therefore these data are not portrayed in Analysis 2.5. The authors reported that the results favoured combined therapy, but the difference was not statistically significant (median change in combined group 32% (interquartile range (IQR) 8.5 to 60.5%), median change in systemic group 18% (IQR 2.0 to 50.5%)).
Frequency‐specific changes with pure tone audiometry
Four studies assessed changes in hearing level at specific frequencies (Ahn 2008; Arslan 2011; Gundogan 2013; Lim 2013). Due to the different frequencies assessed in the studies, and heterogeneity in the effect estimates where studies did assess the same frequency, we did not meta‐analyse these data. Instead the results from each study are shown in Analysis 2.6.
Mean level of improvement, in those whose hearing is improved
This outcome was not reported by any of the included studies.
Percentage of patients reaching serviceable hearing
This outcome was not reported by any of the included studies.
Effect on tinnitus and vertigo
This outcome was not reported by any of the included studies.
Minor and serious adverse events
For this comparison, eight studies provided information pertaining to adverse events (Ahn 2008; Arastou 2013; Arslan 2011; Battaglia 2008; Choi 2011; Gundogan 2013; Koltsidopoulos 2013; Tsounis 2018). In three studies reporting was incomplete for one or more adverse event outcomes, because it was unclear in which group (or groups) events were observed (Arastou 2013; Arslan 2011; Tsounis 2018). There were insufficient data for meta‐analysis.
Persistent tympanic membrane perforation
Five studies reported a rate of perforation between 0% (0/85) and 5.5% (2/36) for those who received an intratympanic injection (Ahn 2008; Arastou 2013; Arslan 2011; Choi 2011; Gundogan 2013). We concluded that the evidence is very uncertain regarding the risk of tympanic membrane perforation for those who received intratympanic corticosteroid combined with systemic corticosteroid as primary treatment (474 participants; 5 studies; very low‐certainty evidence).
Vertigo/dizziness at the time of injection
Four studies reported a rate between 0% (0/60) and 8.1% (3/37) for those who received an intratympanic injection (Ahn 2008; Choi 2011; Gundogan 2013; Tsounis 2018). Note that in one study, two groups received intratympanic injection: one group received intratympanic corticosteroid, and the other received intratympanic and systemic corticosteroid (Tsounis 2018). We concluded that the evidence is very uncertain regarding the risk of vertigo/dizziness at the time of intratympanic injection for those who received intratympanic corticosteroid as primary treatment (341 participants; 4 studies; very low‐certainty evidence).
Ear pain at the time of injection
One study reported a rate of 5/37 (13.5%) for those who received an intratympanic corticosteroid injection (Gundogan 2013). All recovered within one hour. We concluded that the evidence is very uncertain regarding the risk of ear pain at the time of intratympanic injection for those who received combined treatment as primary treatment (73 participants; 1 study; very low‐certainty evidence).
Table 2 provides details of the more limited data on other reported adverse events.
Intratympanic corticosteroids versus no treatment or versus placebo as secondary therapy
There were five studies comparing the efficacy of a secondary intratympanic corticosteroid with no therapy (Chang 2010; Ho 2004; Lee 2011; Li 2011; Xenellis 2006), and two studies comparing intratympanic corticosteroid with intratympanic placebo (Plontke 2009; Wu 2011).
Change in hearing threshold with pure tone audiometry (pure tone average)
Seven studies (280 participants) were included in the meta‐analysis. The mean change in PTA between baseline and 20 to 60 days (range) after the start of therapy showed more improvement in participants with intratympanic treatment but the point estimate of effect did not exceed the minimally important difference of ‐10 dB (MD ‐9.07 dB, 95% CI ‐11.47 to ‐6.66; 280 participants; 7 studies; I2 = 23%; low‐certainty evidence) (Analysis 3.1). Secondary (rescue) intratympanic therapy may therefore result in a small benefit compared to no treatment or placebo, but it is not clear whether this would be important to patients. The studies Plontke 2009 and Wu 2011 were included in a sensitivity analysis and confirmed the result of our primary analysis (MD ‐5.45 dB, 95% CI ‐9.30 to ‐1.59; 76 participants; 2 studies; I2 = 0%).
Proportion of patients whose hearing is improved
Six studies (232 participants) were included in the meta‐analysis. The study Chang 2010 could not be included since this parameter was not reported. A higher proportion of participants with intratympanic therapy had improved hearing at 29 to 79 days (range) after the start of treatment (RR 5.55, 95% CI 2.89 to 10.68; 232 participants; 6 studies; I2 = 0%; low‐certainty evidence) (Analysis 3.2). Secondary (rescue) intratympanic therapy may therefore result in a much higher proportion of patients whose hearing is improved, compared to no treatment or placebo. The studies Plontke 2009 and Wu 2011 were included in the sensitivity analysis and confirmed this result (RR 4.21, 95% CI 1.44 to 12.31; 76 participants; 2 studies; I2 = 0%).
Final hearing threshold with pure tone audiometry (pure tone average)
Five studies (203 participants) were included in the meta‐analysis. The studies Ho 2004 and Chang 2010 could not be included since they did not report a final PTA. The mean final PTA at 29 to 61 days (range) after the start of therapy was lower (better) in participants with intratympanic therapy, and the point estimate of effect exceeded the minimally important difference of ‐10 dB HL (MD ‐11.09, 95% CI ‐17.46 to ‐4.72; 203 participants; 5 studies; I2 = 0%; low‐certainty evidence) (Analysis 3.3). Secondary intratympanic therapy may result, therefore, in lower (improved) final hearing thresholds compared to no treatment or placebo. The studies Plontke 2009 and Wu 2011 were included in the sensitivity analysis and confirmed the result of the primary analysis (MD ‐10.20 dB, 95% CI ‐19.64 to ‐0.77; 76 participants; 2 studies; I2 = 0%).
Change in hearing threshold with speech audiometry
One study reported on this outcome (Plontke 2009). The speech reception threshold may be lower (better) in the group who received intratympanic treatment compared to those who received placebo, although the confidence interval crosses unity (MD ‐12.80 dB, 95% CI ‐30.17 to 4.57; 21 participants; 1 study; Analysis 3.4).
The same study also reported on the change in maximum speech discrimination, measured as the number of monosyllables understood. Again, this outcome appeared to favour intratympanic treatment (Analysis 3.5).
Frequency‐specific changes with pure tone audiometry
A single study reported on frequency‐specific changes with pure tone audiometry. The results are shown in Analysis 3.6.
Mean level of improvement, in those whose hearing is improved
This outcome was not reported by any of the included studies.
Percentage of patients reaching serviceable hearing
This outcome was not reported by any of the included studies.
Effect on tinnitus and vertigo
This outcome was not reported by any of the included studies.
Minor and serious adverse events
For this comparison, five studies provided information about adverse events (Ho 2004; Li 2011; Plontke 2009; Wu 2011; Xenellis 2006). There were insufficient data for meta‐analysis. In all studies, reporting was incomplete for one or more adverse event outcomes, either because a rate was not provided for both randomised groups, or because it was unclear in which group (or groups) events were observed.
Persistent tympanic membrane perforation
Five studies reported a rate of tympanic membrane perforation of between 0% (0/19) and 4.2% (1/24) for those who received an intratympanic injection (Ho 2004; Li 2011; Plontke 2009; Wu 2011; Xenellis 2006). This includes participants who received placebo intratympanic injection. We concluded that the evidence is very uncertain regarding the risk of tympanic membrane perforation for those who received intratympanic injection (either corticosteroid or placebo) as secondary treatment (185 participants; 5 studies; very low‐certainty evidence).
Vertigo/dizziness at the time of intratympanic injection
Three studies reported a rate of vertigo/dizziness of between 6.7% (1/15) and 33% (number not reported) for those who received an intratympanic injection (Ho 2004; Li 2011; Wu 2011). This includes participants who received placebo intratympanic injection. We concluded that the evidence is very uncertain regarding the risk of vertigo/dizziness at the time of intratympanic injection (either corticosteroid or placebo) as secondary treatment (118 participants; 3 studies; very low‐certainty evidence).
Ear pain at the time of intratympanic injection
One study reported no participants with ear pain at the time of intratympanic injection (0/24) (Li 2011). The evidence is very uncertain regarding the risk of ear pain at the time of intratympanic corticosteroid injection as secondary treatment (44 participants; 1 study; very low‐certainty evidence).
Table 3 provides details of the more limited data on other reported adverse events.
Intratympanic corticosteroids versus systemic corticosteroids as secondary therapy
No study compared the effects of intratympanic corticosteroids versus systemic corticosteroids on hearing improvement for secondary therapy of ISSNHL.
Intratympanic plus systemic corticosteroids (combined therapy) versus no treatment or versus placebo as secondary therapy
No study compared the effects of intratympanic corticosteroids plus systemic corticosteroids (combined therapy) versus no treatment or versus placebo on hearing improvement for secondary therapy of ISSNHL.
Intratympanic plus systemic corticosteroids (combined therapy) versus systemic corticosteroids alone as secondary therapy
One study compared the effects of a secondary systemic versus a secondary combined intratympanic and systemic corticosteroid treatment (Zhou 2011).
Change in hearing threshold with pure tone audiometry (pure tone average)
Change in hearing threshold (PTA change) was not reported in the study Zhou 2011.
Proportion of patients whose hearing is improved
One study (76 participants) explored this outcome. A higher proportion of participants with combined therapy had improved hearing at 56 days after the start of treatment compared with the systemic treatment group. The point estimate of effect exceeded the minimally important difference of 25% (RR 2.24, 95% CI 1.10 to 4.55; 76 participants; 1 study; very low‐certainty evidence) (Analysis 4.1). Secondary combined therapy may therefore increase the proportion of patients whose hearing is improved compared to systemic corticosteroids alone, but the evidence is very uncertain.
Final hearing threshold with pure tone audiometry (pure tone average)
This outcome was not reported by Zhou 2011.
Change in hearing threshold with speech audiometry
Zhou 2011 did not assess hearing thresholds, but did report the proportion of participants who achieved an improvement of at least 15% in their speech discrimination score (see Analysis 4.2).
Change in speech discrimination scores
This outcome was not reported by Zhou 2011.
Frequency‐specific changes with pure tone audiometry
This outcome was not reported by Zhou 2011.
Mean level of improvement, in those whose hearing is improved
This outcome was not reported by Zhou 2011.
Percentage of patients reaching serviceable hearing
This outcome was not reported by Zhou 2011.
Effect on tinnitus and vertigo
This outcome was not reported by Zhou 2011.
Minor and serious adverse events
Zhou 2011 provided data for this comparison.
Persistent tympanic membrane perforation
The rate of tympanic membrane perforation in the intervention group was 8.1% (3/37). We concluded that the risk of tympanic membrane perforation among those who receive intratympanic corticosteroid combined with systemic corticosteroid as primary treatment is very uncertain (76 participants; 1 study; very low‐certainty evidence).
Table 4 provides details of the more limited data on other reported adverse events.
Discussion
Summary of main results
We identified data for four of our proposed comparisons, from a total of 30 randomised controlled trials (RCTs) that analysed 2133 participants. No data were found for the comparisons of intratympanic corticosteroids versus no treatment/placebo as primary therapy, intratympanic corticosteroids versus systemic corticosteroids as secondary therapy, or intratympanic corticosteroids plus systemic corticosteroids versus placebo/no treatment as either primary or secondary therapy.
The following is a summary of the key findings for each comparison:
Intratympanic corticosteroids versus systemic corticosteroids as primary therapy
We identified 16 studies and analysed 1108 patients for this comparison (summary of findings Table 1). Intratympanic corticosteroids may result in a trivial or no difference in the change in hearing threshold, as compared with systemic steroids (low‐certainty evidence). They probably also result in little to no difference in the number of participants whose hearing improves, and may result in little to no difference in the final hearing threshold. The confidence intervals of all outcomes do not overlap the thresholds for clinical relevance, and these results persisted after sensitivity analysis. Overall, vertigo and dizziness may be increased, and ear pain is probably more common for those who receive intratympanic corticosteroids. However, adverse effects commonly associated with steroid use (such as blood glucose problems) may be reduced among those who receive intratympanic steroids. Persistent tympanic membrane perforation, ear pain at the time of the injection and vertigo/dizziness at the time of the injection were noted among those who received intratympanic injection, but we could not be certain how often these effects would occur.
Intratympanic plus systemic corticosteroids (combined therapy) versus systemic corticosteroids alone as primary therapy
We identified 10 studies and analysed 788 patients for this comparison (summary of findings Table 2). The change in hearing threshold may be slightly increased (better) among those who received combined therapy, but it is unclear whether this increase would be noticeable and important to patients (low‐certainty evidence). The evidence regarding the number of patients whose hearing improved, and the final hearing threshold, was very uncertain, although both outcomes favoured the combined treatment group. Adverse effects were only reported for those who received combined therapy, therefore we were unable to compare the intervention to systemic corticosteroids. Persistent tympanic membrane perforation, vertigo/dizziness at the time of the injection and ear pain at the time of the injection were all reported in the intervention group, but we could not be certain how often these effects would be seen.
Intratympanic corticosteroids versus no treatment or versus placebo as secondary therapy
Seven studies were included for this comparison (summary of findings Table 3). Five studies compared the efficacy of a secondary intratympanic corticosteroid with no therapy and two studies compared intratympanic corticosteroid with intratympanic placebo. Intratympanic therapy may result in a small improvement in the change in hearing threshold (low‐certainty evidence) although the mean difference was just below the threshold for clinical relevance (PTA change ‐9.07 dB). In addition, intratympanic corticosteroids may result in a much higher proportion of patients achieving an improvement in their hearing (absolute effect of 315 more patients per 1000 having improved hearing) and a small, but clinically important, effect on the final hearing threshold (decrease of 11.09 dB HL, low‐certainty). The clinically relevant effects persisted after sensitivity analysis. Adverse effects were only reported for those who received intratympanic corticosteroids, therefore we were unable to compare the intervention to placebo or no treatment. Persistent tympanic membrane perforation, vertigo/dizziness at the time of the injection and ear pain at the time of the injection were all reported in the intervention group, but we could not be certain how often these effects would be seen. The results indicate that there may be a small improvement in hearing with the use of intratympanic corticosteroids, but it is unclear whether this would be a noticeable or important difference.
Intratympanic plus systemic corticosteroids (combined therapy) versus systemic corticosteroids alone as secondary therapy
We identified one study and analysed 76 patients for this comparison (summary of findings Table 4). Combined therapy may increase the proportion of patients whose hearing is improved, but the evidence is very uncertain. No data were available for the remaining efficacy outcomes for this comparison (change in hearing threshold or final hearing threshold, determined by PTA). The study did report that a number of participants had a persistent tympanic membrane perforation, but the evidence was very uncertain.
Overall completeness and applicability of evidence
The available evidence included all corticosteroids known to be used for intratympanic applications. All studies only included participants with sudden idiopathic sensorineural hearing loss (without alternative diagnoses) and most studies included adults. As participants included in the studies were predominantly adults, it is not clear whether these results also apply to children. However, ISSNHL in children is rare. The available evidence included patients mostly treated in secondary and tertiary care settings. In summary, we conclude that the ISSNHL patients included in the review cover the patient population seen in clinical practice.
The included studies showed variability in their treatment protocols. The glucocorticoids used in the included RCTs were either methylprednisolone or dexamethasone preparations. The methylprednisolone concentrations used were usually 40 mg/mL with only very few studies using higher concentrations (62.5 mg/mL or 125 mg/mL). The dexamethasone concentrations used were usually 4 mg/mL to 5 mg/mL with only one study using a higher concentration (12 mg/mL). We are aware that in clinical practice, other (higher) concentrations of corticosteroids might be used or recommended (Chandrasekhar 2019). In addition, other types of corticosteroids (e.g. triamcinolone acetonide), other forms of corticosteroids (e.g. dexamethasone phosphate versus dexamethasone base) and other drug delivery systems different from intratympanic injections of solutions, including wicks, gels, catheters or biodegradable controlled‐release implants, may be used (reviewed in: El Kechai 2015; Mäder 2018; Salt 2009; Salt 2018; Zhang 2021). However, we did not find any further RCTs that have addressed therapeutic strategies using intratympanic application of corticosteroids for ISSNHL, other than those included in this systematic review.
Three outcome parameters for the evaluation of the treatment effect could be used for statistical analysis (the primary outcome parameter and two secondary outcome parameters). The primary outcome parameter (change in pure tone average (PTA)) is widely used in studies on the treatment of ISSNHL. The proportion of patients whose hearing is improved (one of the secondary outcome parameters) is also a widely used outcome parameter in studies. However, in the view of the authors, this type of outcome parameter is not very reliable because 1) it is a dichotomous parameter, which gives only a little information about the absolute hearing improvement in patients and within‐study group variance and 2) it is highly dependent on the definition of hearing improvement, which is inconsistent between studies. As a further outcome parameter we included the final PTA in patients at the study endpoint. Although this outcome parameter is not widely used in studies, it has been shown to depend less upon baseline characteristics (such as initial hearing loss or treatment delay), and is therefore more robust against distortions due to differences in baseline characteristics between treatment arms (Liebau 2017; Liebau 2018).
In this review, the wording of the comments in the summary of findings tables and, thus, in the abstract, results and authors’ conclusions sections is based on the "GRADE guidelines informative statements to communicate the findings of systematic reviews of interventions" (Santesso 2020). In this guideline, producers and users of systematic reviews found statements to communicate findings that combine the size and certainty of an effect to be acceptable. The final list of informative statements to communicate the results of systematic reviews combines the effect size (1) large effect, 2) moderate effect, 3) small important effect, 4) trivial, small unimportant effect or no effect) and the certainty of the evidence (high, moderate, low, very low) (Santesso 2020). The Cochrane Handbook for Systematic Reviews of Interventions (Chapter 15.6.4) also suggests using these narrative statements to draw conclusions based on the effect estimate from the meta‐analysis and the certainty of the evidence (Handbook 2021). The clinicians amongst the authors initially tried to define the effect sizes in detail (1) large effect, 2) moderate effect, 3) small important effect, 4) trivial, small, unimportant effect or no effect) for both change in hearing threshold and proportion of patients whose hearing improved. However, we could not agree on a uniform statement for this. Thus, we agreed on a 'minimally important difference' (MID).
Determining a relevant and important change in hearing is challenging. In this review we have taken a change in hearing threshold of 10 dB HL to represent the MID. However, we acknowledge that this may not be universally agreed. The decision to choose 10 dB as a MID was based on the test‐retest reliability of pure tone audiometric measurements, established minimal criteria for improvement in individual patients (Chandrasekhar 2019; Gurgel 2012; Stachler 2012), and on a large RCT on this topic with low bias (Rauch 2011). For dichotomous outcomes (e.g. the proportion of patients with hearing improvement), we used a threshold of 25% or more in RR increase for appreciable benefit as suggested in the GRADE guideline (Guyatt 2011). The 10 dB difference and the 25% criteria were agreed upon by all authors. Many of the mean differences reported in this review were close to this MID, therefore it is uncertain whether the detectable change from the interventions would be of importance to patients.
Some studies used different thresholds to define 'improvement' of hearing. This may result in different conclusions to this review. If, for example, a change of 5 dB HL (or 3 dB, 6 dB or 9 dB) was deemed to be the MID then we would have concluded that some interventions were of more certain benefit. This may also partly explain apparent discrepancies in our findings where the mean difference was found not to be clinically relevant, and yet a higher proportion of patients 'improved' when assigned to the intervention group. When interpreting the findings, it is important to consider both the mean change in hearing and how many people improved. Although the mean change for the whole group may not be especially strong, there may still be a greater number of people who improve.
The estimation of whether an effect size is "1. large", "2. moderate", "3. small important" or "4. trivial, small unimportant effect or no effect" (Santesso 2020) also depends on the degree of initial hearing loss (i.e. moderate, severe, profound hearing loss) and whether the patients had serviceable hearing before and/or after therapy (Chandrasekhar 2019). For example, a 10 dB change might not be useful in severe or profound hearing loss if the patient (or the ear) would remain at a cochlear implant candidate level after therapy. The current US guideline therefore correctly recommends that future studies should report the number of patients reaching serviceable hearing: "For ears that were rendered nonserviceable by the episode of SSNHL, return to serviceable hearing should be considered a significant improvement, and whether or not this level of recovery occurs should be recorded. Recovery to a serviceable level typically indicates that after recovery, the ear would be a candidate for traditional hearing amplification. Recovery to less‐than‐serviceable levels indicates an ear that would, in most circumstances, not benefit from traditional amplification. For ears with SSNHL to hearing levels that are still in the serviceable range, an improvement of > 10 dB in pure tone thresholds (accounting for test‐retest variability in audiometry) or an improvement in WRS of > 10% (approximate lower limit for a statistically significant change based on binomial tables for WRS of >50% at baseline) should be considered partial recovery and recorded." (Chandrasekhar 2019).
We think that the criteria with high patient relevance are: how many patients (ears) reach levels where they are not a cochlear implant candidate anymore ("serviceable hearing", as stated above) or even reach levels where a hearing aid would not be necessary any more? Such criteria mainly depend on word recognition tests, which were not sufficiently reported in the RCTs in our review. However, speech audiometry results are difficult to compare due to different test strategies and different languages. The criteria for candidacy for a cochlear implant or a hearing aid may also differ between countries or even between audiologists.
Data for many of the outcomes were missing. The length of follow‐up in studies was less than a year, meaning that there was limited evidence regarding the long‐term effectiveness of the therapies. However, a stable hearing threshold is considered to occur several weeks after treatment of ISSNHL and long‐term follow‐up may increase the likelihood of occurrence of other causes of hearing loss, which would confound any long‐term analysis.
Important or key outcome criteria missing in this review are quality of life and patient‐reported outcome measures. These measures should ideally have been defined in the protocol version of this review. Possible tools for measuring quality of life or patient‐reported outcomes might be the Hearing Handicap Inventory for the Elderly (Ventry 1982), the Short Form (12) Health Survey (SF‐12) (Jenkinson 1997), or the Core Rehabilitation Outcome Set for Single Sided Deafness (CROSSSD) (Katiri 2020). However, quality of life measures had not been reported in any of the RCTs included in this systematic review.
Quality of the evidence
We largely assessed the certainty of the evidence in this review as low or very low. There was moderate‐certainty evidence for a small number of outcomes, but we identified no high‐certainty evidence. The main reasons for the uncertainty were a serious risk of bias in the included studies and imprecision in the effect estimates ‐ either due to a small number of included participants or few events, or because the wide confidence intervals overlapped the threshold for clinical relevance.
Nearly all the included studies had a small number of participants. This increases the risk that randomisation does not achieve balance across groups for important prognostic characteristics (both known and unknown) that may confound outcome estimates. One such characteristic is the propensity towards spontaneous recovery. This can have a large impact on pooled estimates for hearing threshold, and imbalance across groups may not be detected by a comparison of baseline parameters such as pre‐treatment hearing loss and delay between the onset of symptoms and the start of treatment. It is important, therefore, to make sure that trials have enough participants to achieve a balance across groups for the propensity towards spontaneous recovery, as well as other important prognostic factors. This is supported by meta‐analyses (Liebau 2017; Liebau 2018), which demonstrated (in primary and secondary treatment respectively) that variation in outcome estimates is reduced when there is a larger number of participants.
These observations are reflected in the meta‐analysis as well. In many included studies a high within‐group variance was seen that might be the consequence of the heterogeneous impact of spontaneous recovery on total hearing improvement among patients, in combination with a low number of included participants per treatment arm. In addition, heterogeneous results were found in outcome parameters between studies, especially in studies with a small number of included participants. However, due to the high imprecision of the results in these studies statistical heterogeneity might be underestimated by the Chi² test and I² statistic.
We created funnel plots for outcomes including 10 or more studies (Figure 4; Figure 5; Figure 6). These did not indicate the presence of publication bias. However, this is not proof that no publication bias exists. It is noteworthy that we found some high‐quality studies with larger sample sizes indicating no differences between treatment modalities and smaller studies with higher risk of bias indicating large differences in treatment effects. High‐quality studies with large sample sizes imply more investment of time, work and money. It is very likely that these studies will be published afterwards. In smaller studies, a higher risk exists that the results of those studies will not be published if the conclusion differs from current concepts or expectations.
The overall methodological and reporting quality of the studies was disappointing. This leads to an overall high risk of bias in most of the included studies and reduces the certainty of the evidence. It seems that 20 years after CONSORT (Begg 1996), authors and journals in the field of otolaryngology still do not adhere to these guidelines and most of the publications of RCTs were accepted without fulfilling essential methodological and reporting criteria. Further, only very few included studies were reported to be pre‐registered. Without publishing a pre‐specified study protocol, uncertainty remains about whether study methods may have changed after data synthesis and before publishing the study results.
Many studies did not report their method of randomisation. It was therefore not possible to justify whether an adequate randomisation method was used in these studies. Similarly, almost no study reported their method for concealment of allocation. However, the randomisation principle is only guaranteed if ‐ in addition to a robust method of randomisation ‐ a plausible method was used that prevents manipulation of the allocation process. Without this information the randomisation procedure remains questionable. In addition, some studies reported an inadequate method of randomisation in which the allocation of patients could be predicted (such as alternate allocation).
It is also mandatory to give a clear statement about which persons in the study are blinded and which are not. Blinding of outcome assessment is possible even in non‐placebo‐controlled studies. Without a clear statement of blinding a judgement on the risk of bias is not possible and remains unclear.
In some of the included studies, treatment arms actually differed considerably in their baseline parameters and these imbalances between groups may have influenced the reported outcomes. In addition, many studies have not reported important baseline parameters, or reported them in an inadequate way. In two studies the reporting of baseline parameters even differed between treatment arms. Such inconsistent reporting of data raises doubts about the transparency of studies. We also noted a number of studies where measures of variance (such as standard deviations or standard errors) were missing. A variance element is an essential component when reporting the mean.
Potential biases in the review process
The included studies used a broad range of different treatment protocols and follow‐up times for final hearing evaluation. No specific conclusions can thus be drawn on the effectiveness of a particular corticosteroid treatment protocol for ISSNHL. However, this only marginally influences the general conclusions on the comparison of treatment modalities (i.e. systemic, intratympanic or combined treatment).
The included studies also cover a broad range of baseline characteristics (e.g. degree of hearing loss, treatment delay, type of primary treatment for comparisons using intratympanic corticosteroids as secondary treatments, and accompanying symptoms such as vertigo).
There are differences between studies concerning hearing evaluation (different frequencies used for calculating PTA) and different definitions of positive response to treatment (proportion of patients whose hearing is improved). It has previously been shown that the frequencies chosen for calculating the PTA influence the estimation of hearing loss in patients and thus can influence the outcome in clinical trials (Plontke 2007). Different definitions of positive response to treatment can present a risk of bias when pooling the outcome parameter 'proportion of patients improved' (Haynes 2007).
There are many included studies with a high risk of bias and major methodological weaknesses. We therefore performed a sensitivity analysis, which takes into account only studies with high quality and low risk of bias.
One included study was performed by authors of the review (SKP and CM) (Plontke 2009). These authors were therefore not involved in data extraction or risk of bias assessment for this study.
The outcome parameter 'final PTA' after treatment was added after the publication of the protocol. This was based on conclusions drawn from a different meta‐analysis (Liebau 2017; Liebau 2018), and was independent from the data processing in the present meta‐analysis.
Some of the planned secondary outcomes could not be assessed due to the lack of available data. Due to missing individual patient data in almost all studies and the limited number of studies per type of comparison, intended subgroup analyses could not be performed either. However, this has no influence on the conclusions drawn from the overall analysis.
Agreements and disagreements with other studies or reviews
A number of meta‐analyses on local corticosteroid treatment of ISSNHL have already been published. There is much variation concerning included studies, defined types of comparisons and evaluated outcome parameters within these reviews. However, the primary outcome parameter was either mean hearing gain (change in hearing threshold with pure tone audiometry) or recovery rate (proportion of patients whose hearing is improved). Many studies included both outcome parameters.
Intratympanic corticosteroids versus no treatment or versus placebo as primary therapy
We identified no data of relevance for this comparison because of the lack of studies that fulfilled the inclusion criteria. Despite that, the meta‐analysis Ahmadzai 2019 reports a significant benefit of intratympanic treatment over placebo in hearing gain and recovery rate. They did not perform pair‐wise comparisons in their meta‐analysis but pooled single treatment arms from different trials instead. Besides inclusion of the study Filipo 2013 with an early salvage therapy at day seven, they further included the placebo groups from the studies Hultcrantz 2014 and Nosrati‐Zarenoe 2012, although these studies compared the efficacy of systemic corticosteroids versus placebo instead of local treatment. This may lead to a greater risk of bias because effect evaluation is not based on a randomised process.
Intratympanic corticosteroids versus systemic corticosteroids as primary therapy
A large number of meta‐analyses have examined the difference in effectiveness of intratympanic and systemic treatment as primary therapy. Most of them are in accordance with our finding of no important difference between the two treatment modalities (Ahmadzai 2019; Crane 2015; El Sabbagh 2016; Garavello 2012; Lai 2017; Mirian 2020). However, the meta‐analysis Crane 2015 included studies by Ahn 2008 and Arslan 2011, which compared combined rather than local treatment. The review Garavello 2012 included the study Ahn 2008 in this comparison. The meta‐analysis El Sabbagh 2016 did not separate conditions of primary and secondary therapy and pooled studies of both types of treatment. They also included studies with combined treatment in the intervention group in this type of comparison. The meta‐analyses Mirian 2020 and Lai 2017 did not include some of the trials included in our analysis. The meta‐analysis Ahmadzai 2019 pooled single treatment arms from different studies and therefore included the systemic treatment groups of Eftekharian 2016, Gundogan 2013 and Hultcrantz 2014, as well as the local treatment group from Filipo 2013 with early salvage therapy in this comparison.
The meta‐analysis Quiang 2016 found a significantly greater benefit of intratympanic therapy over systemic therapy as a primary intervention when assessed with mean hearing gain as well as recovery rate. This meta‐analysis, however, did not include the large clinical trial with low bias Rauch 2011, in which there were a high number of included participants finding no significant difference in outcomes between treatment modalities. The meta‐analysis Zhao 2016 found a significant difference in the rate of complete hearing recovery, but not in the general recovery rate, termed as significant hearing improvement. However, this review included the studies Mao 2005, Yi 2011, You 2008 and Zhou 2006. The studies Yi 2011 and Mao 2005 investigated the injection of corticosteroids through the Eustachian tube. You 2008 is a non‐randomised study, while Zhou 2006 included hyperbaric oxygen therapy in treatment arms. The meta‐analysis Li 2020 reported a significant difference in mean hearing gain but not in recovery rate. However, they did not include a number of trials that were included in our analysis.
Intratympanic plus systemic corticosteroids (combined therapy) versus systemic corticosteroids alone as primary therapy
Other meta‐analyses about the effectiveness of combined therapy over systemic corticosteroids as primary therapy found heterogeneous results. This is consistent with our conclusion that there is uncertainty in the evidence. Ahmadzai 2019 and Mirian 2020 concluded that combined therapy is not superior to systemic therapy as primary treatment. On the other hand, the meta‐analyses Gao 2016, Han 2017 and Li 2020 showed a significant difference in the effectiveness of combined and systemic therapy (favouring combined therapy) and concluded that combined therapy might be superior to systemic therapy. Gao 2016, however, included two non‐randomised trials (Battaglia 2014; Günel 2015). In Günel 2015, the control group was a retrospective cohort. Han 2017 also included the study Ashtiani 2012, a study with a very uncertain randomisation process, which is only mentioned in the abstract, not in the full text of their publication. They also included the non‐randomised trial Battaglia 2014 and the trial Chen 2015, in which the control group included a mixture of patients receiving combined or systemic treatment. Li 2020 also included the non‐randomised trials Battaglia 2014 and Ashtiani 2012 (very uncertain randomisation process). Further, they included the trial Zhou 2011, which investigated the efficacy of combined therapy against systemic treatment as a secondary intervention.
Intratympanic corticosteroids versus no treatment or versus placebo as secondary therapy
In accordance with our meta‐analysis, other reviews also found that intratympanic salvage therapy might be more effective than no therapy or placebo therapy. The meta‐analyses Ng 2015 and Li 2015 found a significantly higher mean change in hearing threshold in patients receiving intratympanic therapy. The meta‐analysis Spear 2011 also found a significant difference in mean change in hearing threshold. However, this review also included non‐randomised studies (Kiliç 2007; Plaza 2007; She 2010). The reviews Crane 2015 and Garavello 2012 found a significantly higher recovery rate in patients receiving intratympanic therapy. The review Crane 2015, however, included the study Zhou 2011, which compared combined treatment with systemic therapy. The review Garavello 2012 also included the study Zhou 2011 and the study Arslan 2011, which compared combined therapy with systemic treatment as primary therapy.
Combined intratympanic plus systemic corticosteroids versus systemic corticosteroids alone as secondary therapy
We found no review addressing this question.
Process of selection of studies for inclusion in the review.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Comparison 1: Intratympanic compared to systemic corticosteroids as primary therapy, Outcome 1: Mean change in pure tone average (PTA)
Comparison 1: Intratympanic compared to systemic corticosteroids as primary therapy, Outcome 2: Proportion improved
Comparison 1: Intratympanic compared to systemic corticosteroids as primary therapy, Outcome 3: Final PTA
Comparison 1: Intratympanic compared to systemic corticosteroids as primary therapy, Outcome 4: Change in hearing threshold with speech audiometry
Comparison 1: Intratympanic compared to systemic corticosteroids as primary therapy, Outcome 5: Speech audiometry: additional outcomes
Comparison 1: Intratympanic compared to systemic corticosteroids as primary therapy, Outcome 6: Frequency‐specific changes in PTA
Comparison 1: Intratympanic compared to systemic corticosteroids as primary therapy, Outcome 7: Adverse events
Comparison 2: Combined compared to systemic corticosteroids as primary therapy, Outcome 1: Mean change in pure tone average (PTA)
Comparison 2: Combined compared to systemic corticosteroids as primary therapy, Outcome 2: Proportion improved
Comparison 2: Combined compared to systemic corticosteroids as primary therapy, Outcome 3: Final PTA
Comparison 2: Combined compared to systemic corticosteroids as primary therapy, Outcome 4: Change in hearing threshold with speech audiometry
Comparison 2: Combined compared to systemic corticosteroids as primary therapy, Outcome 5: Speech audiometry: additional outcomes
Comparison 2: Combined compared to systemic corticosteroids as primary therapy, Outcome 6: Frequency‐specific changes with PTA
Comparison 3: Intratympanic corticosteroids compared to no treatment/placebo as secondary therapy, Outcome 1: Mean change in PTA
Comparison 3: Intratympanic corticosteroids compared to no treatment/placebo as secondary therapy, Outcome 2: Proportion improved
Comparison 3: Intratympanic corticosteroids compared to no treatment/placebo as secondary therapy, Outcome 3: Final PTA
Comparison 3: Intratympanic corticosteroids compared to no treatment/placebo as secondary therapy, Outcome 4: Change in hearing threshold with speech audiometry
Comparison 3: Intratympanic corticosteroids compared to no treatment/placebo as secondary therapy, Outcome 5: Speech audiometry: additional outcomes
Comparison 3: Intratympanic corticosteroids compared to no treatment/placebo as secondary therapy, Outcome 6: Frequency‐specific changes with PTA
Comparison 4: Combined compared to systemic corticosteroids as secondary treatment, Outcome 1: Proportion improved
Comparison 4: Combined compared to systemic corticosteroids as secondary treatment, Outcome 2: Speech audiometry: additional outcomes
| Intratympanic corticosteroids versus systemic corticosteroids as primary therapy | ||||||
| Patient or population: sudden sensorineural hearing loss Settings: initial therapy Intervention: intratympanic steroid therapy Comparison: systemic steroid therapy | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | No of participants | Relative effect | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Systemic therapy (assumed risk) | Intratympanic therapy (corresponding risk) | |||||
| Change in hearing threshold determined by PTA
Range 0 dB to 140 dB
Negative values represent lowering and positive values represent raising of the hearing threshold. A lower hearing threshold represents hearing improvement).
| The mean change in PTA ranged across control groups from ‐30.07 dB to ‐15.1 dB | The mean change in PTA in the intervention groups was on average ‐5.93 dB greater (from ‐4.26 greater to ‐7.61 greater) | 701 (10 studies) | MD ‐5.93 dB (95% CI ‐7.61 to ‐4.26) | ⊕⊕⊝⊝ | Intratympanic therapy may have a trivial/no effect on the change in hearing threshold when compared to systemic steroids (as primary therapy). |
| Proportion of patients whose hearing is improved | 731 per 1000a | 760 per 1000 (709 to 818) | 972 (14 studies) | RR 1.04 (95% CI 0.97 to 1.12) | ⊕⊕⊕⊝ | Intratympanic therapy probably results in little to no difference in the proportion of patients whose hearing is improved compared to systemic corticosteroids (as primary therapy). |
| Final hearing threshold determined by PTA (a lower value represents better hearing) | The mean final PTA ranged across control groups from 25.1 dB to 59 dB | The mean final PTA in the intervention groups was on average ‐3.31 dB lower (‐6.16 lower to ‐0.47 lower) | 516 (7 studies) | MD ‐3.31 dB (95% CI ‐6.16 to ‐0.47) | ⊕⊕⊝⊝ low3 | Intratympanic therapy may result in little to no difference in the final hearing threshold (as primary therapy). |
| Adverse eventsb | Events in control group | Events in intervention group | No of participants (studies) | Relative effect (95% CI) | Certainty of the evidence (GRADE) | Comments |
| Tympanic membrane perforation | Comparison not applicablec | Ranged from 0% (0/30) to 3.9% (5/129) | 463 (4 studies) | Not calculable | ⊕⊝⊝⊝ very low4 | The evidence is very uncertain regarding the risk of tympanic membrane perforation for those who received intratympanic corticosteroid as primary treatment. |
| Vertigo/dizziness: timing not reportedd | 13/121 (10.7%) | 35/129 (27.1%) | 250 (1 study) | RR 2.53 (1.41 to 4.54) | ⊕⊕⊝⊝ low5 | Intratympanic therapy may increase the risk of vertigo/dizziness of unspecified timing as compared to systemic corticosteroid. |
| Vertigo/dizziness: at the time of injection | Comparison not applicablec | 3 studies reported a rate between 1.5% (1/67) and 21% (4/19) for those who received an intratympanic injectione | 301 (4 studies) | Not calculable | ⊕⊝⊝⊝ very low6 | The evidence is very uncertain regarding the risk of vertigo/dizziness at the time of intratympanic injection of corticosteroid as primary treatment. |
| Ear pain: timing not reportedf | 4/141 (2.8%) | 74/148 (50%) | 289 (2 studies) | RR 15.68 (95% CI 6.22 to 39.49) | ⊕⊕⊕⊝ | Intratympanic corticosteroid injection probably increases the risk of ear pain of unspecified timing as compared to systemic corticosteroid when used as primary treatment. |
| Ear pain: at the time of injectionf | Comparison not applicablec | 3 studies reported a rate between 4.8% (5/104) and 27.1% (35/129) | 393 (3 studies) | Not calculable | ⊕⊕⊝⊝ | The evidence suggests that there may be a risk of ear pain at the time of intratympanic injection of corticosteroid as primary treatment. |
| *The basis for the assumed risk is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| aFourteen studies recruited participants suffering from sudden sensorineural hearing loss. The incidence of improvement for the systemic corticosteroid group in these 14 studies was 73.07%. We have used 731 per 1000 to express the assumed risk. bOnly the most widely reported adverse events are described here. For adverse events that could feasibly occur in either group, we have only included the studies that provided a rate for both groups. For adverse events that could only occur in one group, we have only included the studies that reported the rate in that group, and presented these as a range. A full description of adverse event data is available for reference in Table 1. cComparisons between patients receiving intratympanic therapy and those receiving only systemic therapy were regarded as invalid for the following adverse events: persistent tympanic membrane perforation, vertigo observed at the time of intratympanic injection and ear pain observed at the time of intratympanic injection. This is explained in Data extraction and management. dA single study reported a rate for both intratympanic and systemic corticosteroid (Rauch 2011). However, it is not specified whether all of the patients in the intratympanic corticosteroid group experiencing vertigo did so at the time of injection. We have therefore reported this outcome separately from vertigo/dizziness interpreted as having occurred specifically at the time of injection. eIn two studies, two groups received intratympanic injection: in Tsounis 2018, one group received intratympanic corticosteroid and the other received intratympanic and systemic corticosteroid; in Huang 2021, one group received intratympanic corticosteroid and the other received intravenous followed by intratympanic corticosteroid. fIn each study contributing data, the number of participants with ear pain/earache was presented separately from the numbers with ear pain at intratympanic injection. It was assumed, therefore that those participants with pain at injection were not included among those with ear pain/earache.
GRADE Working Group grades of evidence | ||||||
| 1Downgraded one level due to risk of bias: eight studies were at high risk of other bias, three studies were at risk of attrition bias and three studies were at risk of selection bias. Downgraded one level due to inconsistency: the size and direction of effect varied between the studies and the I2 value was 80%. 2Downgraded one level due to risk of bias: we judged 11 of 14 studies to be at unclear or high risk of selection bias and we judged 12 of 14 studies to be at high risk of other bias. 3Downgraded one level due to risk of bias: we judged six studies to be at high risk of other bias; two studies were at high risk of selection bias. Downgraded one level due to inconsistency: the I2 value was moderate (41%). 4Downgraded one level due to risk of bias: we judged one study to be at high risk of bias because of concern about random sequence generation and allocation concealment. Downgraded two levels due to imprecision: the number of events is smaller than the optimal information size (taken as 300 events) and an effect estimate could not be calculated. 5Downgraded one level due to imprecision: the number of events is smaller than the optimal information size (taken as 300 events). Downgraded one level because of indirectness: provision of data by only a single study from a single setting, which may not adequately represent all patients with ISSNHL. 6Downgraded two levels due to risk of bias: we judged two studies to be at high risk of bias because of incomplete outcome data; we judged one study to be at high risk of bias because of concern about random sequence generation and allocation concealment. Downgraded two levels due to imprecision: the number of events is smaller than the optimal information size (taken as 300 events) and an effect estimate could not be calculated. 7Downgraded one level due to imprecision: the number of events is smaller than the optimal information size (taken as 300 events). 8Downgraded two levels because of imprecision: the number of events is smaller than the optimal information size (taken as 300 events) and an effect estimate could not be calculated. | ||||||
| Adverse event reported | Study | How reported | Details of recovery | Rate in intervention group (%) | Rate in comparator group (%) | RR (95% CI) |
|---|---|---|---|---|---|---|
| Tympanic membrane (TM) perforation | There were no cases of [...] perforation of the tympanic membrane | NA | 0/52 (0) | 0/52 (0)* | NA | |
|
| No residual TM perforations | All patients demonstrated a complete healing of TM after the tympanostomy tube removal | 0/24 (0) | NA | NA | |
|
| Persistent TM perforation | By the 6‐month follow‐up most adverse events had resolved | 5/129 (3.9) | NA | NA | |
|
| No residual tympanic membrane perforations were observed in any of the individuals at their final visit | NA | 0/30 (0) | NA | NA | |
| Vertigo/dizziness | Four patients in the ITS group had transient vertigo during the procedure | NR | 4/19 (21) | NA | NA | |
|
| [...] complained of brief dizziness after IT injection [...]
| No patients stopped the treatment | 7/52 (13.5) | 8/52 (15.4)* | NA | |
|
| The intratympanic group experienced adverse effects typical of local injection, most often transient pain at the injection site and brief caloric vertigo. Note, it is unclear whether all reported instances of vertigo in the intervention group occurred at the time of injection. | By the 6‐month follow‐up most adverse events had resolved | 35/129 (27.1) | 13/121 (10.7) | 2.53 (1.41 to 4.54); favours systemic corticosteroid; P = 0.002 | |
|
| Temporary adverse events in 22.7% of patients treated with oral prednisolone which included [...] and dizziness. In 35% of patients treated with intratympanic corticosteroid, adverse events occurred including [...] and dizziness. | NR | NR | NR | NA | |
|
| Six of 30 patients in the intratympanic injection group complained of a transient dizziness lasting about a minute during treatment | NR | 6/30 (20) | NA | NA | |
|
| One patient experienced transient dizziness as a result of caloric stimulation from the injected steroid solution (unclear which of 2 groups receiving intratympanic injection) | Symptoms resolved completely within 15 minutes and there was no need to discontinue the treatment. The injections that followed caused no further side effect. | NR | NR | NA | |
| Tinnitus | In 35% of patients treated with intratympanic corticosteroid, adverse events occurred including [...] ringing sensation in the ear | NR | NR | NR | NA | |
| Ear pain | Pain due to injection | NR | 2/19 (10.5) | NA | NA | |
|
|
| Earache | NR | 4/19 (21.1) | 0/20 (0) | 9.45 (0.54 to 164.49); favours systemic corticosteroid; P = 0.12 |
|
| [...] refused repeated IT injections due to unbearable pain | NR | 3/52 (5.8) | 2/52 (3.8) | NA | |
|
| The intratympanic group experienced adverse effects typical of local injection, most often transient pain at the injection site [...] | By the 6‐month follow‐up most adverse events had resolved | 35/129 (27.1) | NA | NA | |
|
|
| Experienced ear pain at least once | By the 6‐month follow‐up most adverse events had resolved | 70/129 (54.3) | 4/121 (3.3) | 16.41 (6.18 to 43.59); favours systemic corticosteroid; P < 0.00001 |
|
| In 35% of patients treated with intratympanic corticosteroid adverse events occurred including […] mild ear pain, severe ear pain (3 patients) | NR | 3/20 (15) severe ear pain | NR | NA | |
|
| Some patients had a tolerable pain reaction after the injection | NR | NR | NA | NA | |
| Other | Mood change | NR | 2/19 (10.5) | 8/20 (40) | 0.26 (0.06 to 1.08); favours IT corticosteroid; P = 0.06 | |
|
|
| Blood glucose problem | NR | 3/19 (15.8) | 6/20 (30) | 0.53 (0.15 to 1.81); favours IT corticosteroid; P = 0.31 |
|
|
| Sleep change | NR | 1/19 (5.3) | 6/20 (30) | 0.18 (0.02 to 1.32); favours IT corticosteroid; P = 0.09 |
|
|
| Increased appetite | NR | 1/19 (5.3) | 5/20 (25) | 0.21 (0.03 to 1.64); favours IT corticosteroid; P = 0.14 |
|
|
| Mouth dryness/thirst | NR | 0/19 (0) | 5/20 (25) | 0.10 (0.01 to 1.62); favours IT corticosteroid; P = 0.10 |
|
|
| Weight gain | NR | 0/19 (0) | 3/20 (15) | 0.15 (0.01 to 2.72); favours IT corticosteroid; P = 0.20 |
|
| No complications related to the treatment were noted in both the groups | NA | 0/25 (0) | 0/21 (0) | NA | |
|
| No long‐term complications were observed in any of the patients | NA | 0/19 (0) | 0/16 (0) | NA | |
|
| No side effects were observed in either group | NA | 0/32 (0) | 0/31 (0) | NA | |
|
| Apparent bleeding at intratympanic injection site | NR | 0/52 (0) | 0/52 (0) | NA | |
|
|
| External otitis or myringitis | NR | 0/52 (0) | 0/52 (0) | NA |
|
|
| Otitis media | NR | 0/52 (0) | 0/52 (0) | NA |
|
|
| Fluctuation of basal blood pressure (> 10 mmHg) | NR | 2/52 (3.8) | 7/52 (13.5) | NA |
|
|
| Fluctuation of fasting blood glucose (> 2 mmol/L) | NR | 5/52 (9.6) | 12/52 (23.1) | NA |
|
|
| Emotional change | NR | 8/52 (15.4) | 15/52 (28.8) | NA |
|
|
| Appetite change | NR | 13/52 (25.0) | 25/52 (48.1) | NA |
|
|
| Dyssomnia | NR | 23/52 (44.2) | 38/52 (73.1) | NA |
|
|
| Water‐sodium retention | NR | 9/52 (17.3) | 24/52 (46.1) | NA |
|
|
| Acne on face and body | NR | 2/52 (3.8) | 6/52 (11.5) | NA |
|
|
| Irregular menstruation | NR | 5/21 (23.8) | 11/23 (47.8) | NA |
|
|
| Cushing's syndrome | NR | 0/52 (0) | 1/52 (1.9) | NA |
|
|
| Osteoporotic fracture |
| 0 | 0 |
|
|
| In one case an acute suppurative otitis media developed that was eliminated by local antibacterial therapy. This patient was excluded from the study. | NA | NA | NA | NA | |
|
|
| Nine patients in the ST (standard therapy) group and 12 patients in the intravenous corticosteroid group complained of sleep loss | Completely corrected after withdrawal | NR | 9/24 (37.5) in ST group and 12/25 (48) in intravenous corticosteroid group | NA |
|
|
| No systemic adverse effects related to intratympanic application of steroids were noticed | NA | 0/24 (0) | NA | NA |
|
|
| No serious side effects related to systemic administration of steroids were observed in the study | NA | NA | 0/49 (0) | NA |
|
| No complications were seen in patients (unclear which group), including those with hypertension or diabetes | NA | NR | NR | NA | |
|
| Mood change | By the 6‐month follow‐up most adverse events had resolved | 12/129 (9.3) | 54/121 (44.6) | 0.21 (0.12 to 0.37); favours IT corticosteroid; P < 0.00001 | |
|
|
| Blood glucose problem | By the 6‐month follow‐up most adverse events had resolved | 21/129 (16.3) | 36/121 (29.8) | 0.55 (0.34 to 0.88); favours IT corticosteroid; P = 0.01 |
|
|
| Sleep change | By the 6‐month follow‐up most adverse events had resolved | 9/129 (7) | 44/121 (36.4) | 0.19 (0.1 to 0.38); favours IT corticosteroid; P < 0.00001 |
|
|
| Appetite change | By the 6‐month follow‐up most adverse events had resolved | 6/129 (4.7) | 28/121 (23.1) | 0.2 (0.09 to 0.47); favours IT corticosteroid; P = 0.0002 |
|
|
| Dry mouth/thirst | By the 6‐month follow‐up most adverse events had resolved | 5/129 (3.9) | 30/121 (24.8) | 0.16 (0.06 to 0.39); favours IT corticosteroid; P < 0.0001 |
|
|
| Weight change | By the 6‐month follow‐up most adverse events had resolved | 7/129 (5.4) | 22/121 (18.2) | 0.3 (0.13 to 0.67); favours IT corticosteroid; P = 0.004 |
|
|
| Ear infection | By the 6‐month follow‐up most adverse events had resolved | 7/129 (5.4) | 2/121 (1.7) | 3.28 (0.7 to 15.49); favours systemic corticosteroid; P = 0.13 |
|
|
| Any adverse event: "Adverse events were reported by 87.6% (106 of 121) of participants in the oral group and 89.9% (116 of 129) in the intratympanic group." Note: it is unclear whether 'adverse events' refers to those already reported (and listed for this study in this table). | By the 6‐month follow‐up most adverse events had resolved | 116/129 (89.9) | 106/121 (87.6) | 1.03 (0.94 to 1.12); favours systemic corticosteroid; P = 0.56 |
|
|
| Serious adverse events: "In the intratympanic treatment group, these included osteomyelitis of the toe, leukemia, myocardial infarction, bladder cancer, chest pain due to possible endocarditis, and exacerbation of pre‐existing chronic obstructive pulmonary disease. In the oral treatment group, the serious adverse events were myocardial infarction, cerebral hemorrhage, hyponatremia, hospitalization for possible transient ischemic attack, and syncope. The case of hyponatremia arose from worsening of pre‐existent mild renal insufficiency in a patient with type 2 diabetes that was deemed study‐related." | NR | 6/129 (4.7) | 5/121 (4.1) | 1.13 (0.35 to 3.59); favours systemic corticosteroid; P = 0.84 |
|
| No adverse effects were reported in either study group during the study period | NA | NA | NA | NA | |
|
| Temporary adverse events: temporary adverse events in 22.7% of patients treated with oral prednisolone which included puffiness of face, ulcers in mouth, increased appetite, diarrhea [...] | NR | NR | 5/22 (22.7) | NA | |
|
|
| Adverse events: in 35% of patients treated with intratympanic methylprednisolone adverse events occurred including mild ear pain, severe ear pain (3 patients), ringing sensation in ear and dizziness | NR | 7/20 (35) | NR | NA |
|
| No otitis media […] observed in any of the individuals at their final visit | NA | 0/30 (0) | 0/30 (0) | NA | |
|
| No significant complications during the intratympanic injections or the follow‐up period | NA | 0/33 (0) | NA | NA | |
| *Patients in the comparator group for this trial also received intratympanic (IT) corticosteroid at a later time point in the trial, therefore complications of IT treatment are included here (Huang 2021). The intervention group received 24 days of IT corticosteroid and the comparator group received 12 days of systemic (intravenous corticosteroid) followed by 12 days of IT corticosteroid. Rate ratios are not presented as they are not applicable to the comparison of interest (IT compared to systemic as primary therapy). CI: confidence interval; IT: intratympanic; NA: not applicable; NR: not reported; RR: risk ratio; TM: tympanic membrane | ||||||
| Intratympanic plus systemic corticosteroids (combined therapy) versus systemic corticosteroids alone as primary therapy | ||||||
| Patient or population: sudden sensorineural hearing loss Settings: initial therapy Intervention: combination of intratympanic and systemic steroid therapy Comparison: systemic steroid therapy | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | No of participants | Relative effect | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Systemic therapy (assumed risk) | Combined therapy (corresponding risk) | |||||
| Change in hearing threshold determined by PTA Range 0 dB to 140 dB Negative values represent lowering and positive values represent raising of the hearing threshold. A lower hearing threshold represents hearing improvement.
| The mean change in PTA ranged across control groups from | The mean change in PTA in the intervention groups was on average ‐8.55 dB greater (‐4.61 greater to ‐12.48 greater) | 435 (6 studies) | MD ‐8.55 dB (95% CI ‐12.48 to ‐4.61) | ⊕⊕⊝⊝ low1 | The change in hearing threshold may be slightly increased in participants who receive combined therapy. However, it is unclear whether this increase would be noticeable to patients. |
| Proportion of patients whose hearing is improved | 579 per 1000a | 735 per 1000 (666 to 816) | 788 (10 studies) | RR 1.27 (95% CI 1.15 to 1.41) | ⊕⊝⊝⊝ | The evidence is very uncertain as to whether combined therapy changes the proportion of participants whose hearing is improved. |
| Final hearing threshold determined by PTA A lower value represents better hearing | The mean final PTA ranged across control groups from 39.1 dB to 59 dB | The mean final PTA in the intervention groups was on average 9.11 dB lower (1.67 lower to 16.56 lower) | 194 (3 studies) | MD ‐9.11 dB (95% CI ‐16.56 to ‐1.67) | ⊕⊝⊝⊝
| Combined therapy may result in slightly lower (more favourable) final hearing thresholds compared to systemic corticosteroids alone (as primary therapy) but the evidence is very uncertain, and it is not clear whether the change would be of importance to patients. |
| Adverse eventsb | Events in control group | Events in intervention group | No of Participants | Relative effect | Certainty of the evidence | Comments |
| Persistent tympanic membrane perforation | Comparison not applicablec | 5 studies reported a rate between 0% (0/85) and 5.5% (2/36) for those who received an intratympanic injection | 474 (5 studies) | Not calculable | ⊕⊝⊝⊝ very low4 | The evidence is very uncertain regarding the risk of tympanic membrane perforation for those who received intratympanic steroids. |
| Vertigo/dizziness: timing not reported | No study reported on this outcome for both the intervention and comparator groups. | |||||
| Vertigo/dizziness: at the time of injection | Comparison not applicablec | 4 studies reported a rate between 0% (0/60) and 8.1% (3/37) for those who received an intratympanic injectiond | 341 (4 studies) | Not calculable | ⊕⊝⊝⊝ very low5 | The evidence is very uncertain regarding the risk of vertigo/dizziness at the time of intratympanic injection for those who received intratympanic corticosteroid as primary treatment. |
| Ear pain: timing not reported | No study reported on this outcome for both the intervention and comparator groups. | |||||
| Ear pain: at the time of injection | Comparison not applicablec | One study reported a rate of 5/37 (13.5%) | 73 (1 study) | Not calculable | ⊕⊝⊝⊝ very low6 | The evidence is very uncertain regarding the risk of ear pain at the time of intratympanic injection for those who received combined treatment as primary treatment. |
| *The basis for the assumed risk is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| aTen studies recruited participants suffering from sudden sensorineural hearing loss. The incidence of improvement for the 10 studies was 57.86%. We have used 579 per 1000 to express the assumed risk. bOnly the most widely reported adverse events are described here. For adverse events that could feasibly occur in either group, we have only included the studies that provided a rate for both groups. For adverse events that could only occur in one group, we have only included the studies that reported the rate in that group, and presented these as a range. A full description of adverse event data is available for reference in Table 2. cComparisons between patients receiving intratympanic therapy and those receiving only systemic therapy were regarded as invalid for the following adverse events: persistent tympanic membrane perforation, vertigo observed at the time of intratympanic injection and ear pain observed at the time of intratympanic injection. This is explained in Data extraction and management. dIn one study, two groups received intratympanic injection: one group received intratympanic corticosteroid and the other received intratympanic and systemic corticosteroid (Tsounis 2018).
GRADE Working Group grades of evidence | ||||||
| 1Downgraded one level due to risk of bias: we rated a study contributing moderate weight to the overall effect estimate as high risk of bias due to concern about random sequence generation and allocation concealment. Five studies were at high risk of other bias, and one study was at risk of attrition bias. Downgraded one level due to imprecision: the 95% CI overlaps the threshold for clinical relevance, taken to be 10 dB. 2Downgraded one level due to risk of bias: we judged 8 of 10 studies to be at high or unclear risk of selection bias and at high risk of other bias. Downgraded one level due to imprecision: the 95% CI overlaps the threshold for clinical relevance. Downgraded one level due to inconsistency: the I2 value was moderate (47%). 3Downgraded two levels due to risk of bias: we judged all three studies to be at high or unclear risk of selection bias and high risk of other bias. We also judged one of three studies to be at high risk of bias for incomplete outcome data and selective reporting. Downgraded one level due to imprecision: the 95% CI overlaps the threshold for clinical relevance and the sample size is smaller than the optimal information size (taken as 400 participants). 4Downgraded two levels due to risk of bias: we judged two studies to be at high risk of bias because of concern about random sequence generation, two studies because of selective reporting, one study because of concern about blinding and one study because of concern about allocation concealment. 5Downgraded two levels due to risk of bias: we judged one study to be at high risk of bias because of concern about random sequence generation and blinding, one study because of selective reporting and one study because of incomplete outcome data. Downgraded two levels due to imprecision: the number of events is smaller than the optimal information size (taken as 300 events) and an effect estimate could not be calculated. 6Downgraded two levels due to imprecision: the number of events is smaller than the optimal information size (taken as 300 events) and an effect estimate could not be calculated. Downgraded one level because of indirectness: provision of data by only a single study from a single setting, which may not adequately represent all patients with ISSNHL. | ||||||
| Adverse event reported | Study | How reported | Details of recovery | Rate in intervention group (%) | Rate in comparator group (%) | RR (95% CI) |
|---|---|---|---|---|---|---|
| TM perforation | No significant complications during or after IT dexamethasone, including TM perforation […] | NR | 0/60 (0) | NA | NA | |
|
| Two patients developed tympanic perforation (reported as 2.6% of whole study sample; unclear how many from each group) | Treated with cauterisation and paper patch (n = 1) and tympanoplasty (n = 1) | NR | NR | NA | |
|
| None of the patients had an important complication, namely […] TM perforation (unclear which group) | NR | NR | NR | NA | |
|
| No significant complications during or after IT steroid injection including TM perforation | NR | 0/19 (0) | NA | NA | |
|
| No case of residual TM perforation […] was noted | No long‐term complications resulting from either oral steroid or IT steroid in any of the patients | 0/37 (0) | NA | NA | |
| Vertigo/dizziness | No significant complications during or after IT dexamethasone, including […] vertigo […] | NR | 0/60 (0) | NA | NA | |
|
| None of the patients had an important complication, namely, […] vertigo […] (unclear which group) | NR | NR | NR | NA | |
|
| No significant complications during or after IT steroid injection, including […] vertigo […] | NR | 0/19 (0) | NA | NA | |
|
| Three patients complained of vertigo immediately after injection | Recovered after 2 hours of rest | 3/37 (8.1) | NA | NA | |
|
| One patient experienced transient dizziness as a result of caloric stimulation from the injected steroid solution (unclear which of two groups receiving intratympanic injection) | Symptoms resolved completely within 15 minutes and there was no need to discontinue the treatment. The injections that followed caused no further side effects. | NR | NR | NA | |
| Tinnitus | No significant | NR | 0/60 (0) | NA | NA | |
|
| None of the patients had an important complication, namely, […] tinnitus […] (unclear which group) | NR | NR | NR | NA | |
|
| No significant complications during or after IT steroid injection, including […] tinnitus […] | NR | 0/19 (0) | NA | NA | |
| Ear pain | None of the patients had an important complication, namely, […] otalgia […] (unclear which group) | NR | NR | NR | NA | |
|
| Otalgia occurred in 5 patients after IT corticosteroid injection | Relieved after 1 hour | 5/37 (13.5) | NA | NA | |
| Other | No significant complications during or after IT dexamethasone, including […] otitis media […] | NR | 0/60 (0) | NA | NA | |
|
| None of the patients had an important complication, namely, […] nystagmus, otitis media […] (unclear which group) | NR | NR | NR | NA | |
|
| No long‐term complications resulted from either the prednisone taper or the IT corticosteroid in any of the patients enrolled in the study | NR | 0/16 (0) | 0/18 (0) | NA | |
|
| No significant complications during or after IT steroid injection, including […] otitis media […] | NR | 0/19 (0) | NA | NA | |
|
| No case of […] otitis media was noted | No long‐term complications resulted from either oral steroid or IT steroid in any of the patients | 0/37 | NA | NA | |
|
|
| No long‐ term complications resulted from either oral steroid or IT steroid in any of the patients | NA | 0/37 (0) | 0/36 (0) | NA |
|
| No significant complications during the intratympanic injections or the follow‐up period | NA | 0/33 (0) | NA | NA | |
|
| No significant complications occurred during IT injections or the follow‐up period. One case of otitis media was encountered (unclear which group) | NR | 0/46 (0) (significant complications) | NA | NA | |
| IT: intratympanic; NA: not applicable; NR: not reported; TM: tympanic membrane | ||||||
| Intratympanic corticosteroids versus no treatment or versus placebo as secondary therapy | ||||||
| Patient or population: sudden sensorineural hearing loss Settings: after treatment failure with systemic steroids Intervention: intratympanic steroid therapy Comparison: no treatment/placebo | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | No of participants | Relative effect | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| No treatment/placebo (assumed risk) | Intratympanic therapy (corresponding risk) | |||||
| Change in hearing threshold determined by PTA Range 0 dB to 140 dB Negative values represent lowering and positive values represent raising of the hearing threshold. A lower hearing threshold represents hearing improvement.
| The mean change in PTA ranged across control groups from | The mean change in PTA in the intervention groups was on average ‐9.07 dB greater (‐6.66 greater to ‐11.47 greater) | 280 (7 studies) | MD ‐9.07 dB (95% CI ‐11.47 to ‐6.66) | ⊕⊕⊝⊝
| Intratympanic therapy may have a small effect on hearing threshold compared to no treatment or placebo (as secondary therapy), but it is not clear whether this change would be important to patients. |
| Proportion of patients whose hearing is improved | 70 per 1000a | 385 per 1000 (203 to 747) | 232 (6 studies) | RR 5.55 (95% CI 2.89 to 10.68) | ⊕⊕⊝⊝ | Intratympanic therapy may result in a much higher proportion of patients whose hearing is improved, compared to no treatment or placebo (as secondary therapy). |
| Final hearing threshold determined by PTA (a lower value represents better hearing) | The mean final PTA ranged across control groups from 59.9 to 90.5 dB HL | The mean final PTA in the intervention groups was on average ‐11.09 dB lower (‐4.72 lower to ‐17.46 lower) | 203 (5 studies) | MD ‐11.09 dB (95% CI ‐17.46 to ‐4.72) | ⊕⊕⊝⊝ | Intratympanic therapy may result in lower (more favourable) final hearing thresholds compared to no treatment or placebo (as secondary therapy). |
| Adverse eventsb | Events in control group | Events in intervention group | No of Participants (studies) | Relative effect (95% CI) | Certainty of the evidence (GRADE) | Comments |
| Persistent tympanic membrane perforation | Comparison not applicablec | 5 studies reported a rate between 0% (0/19) and 4.2% (1/24) for those who received an intratympanic injectiond | 185 (5 studies) | Not calculable | ⊕⊝⊝⊝ very low4 | The evidence is very uncertain regarding the risk of tympanic membrane perforation for those who received intratympanic injection (either corticosteroid or placebo) as secondary treatment. |
| Vertigo/dizziness: timing not reported | No study reported on this outcome for both the intervention and comparator groups. | |||||
| Vertigo/dizziness at the time of intratympanic injection | Comparison not applicablec | 3 studies reported a rate between 6.7% (1/15) and 33% (number not reported) for those who received an intratympanic injection.d | 128 (3 studies) | Not calculable | ⊕⊝⊝⊝ very low5 | The evidence is very uncertain regarding the risk of vertigo/dizziness at the time of intratympanic injection (either corticosteroid or placebo) as secondary treatment. |
| Ear pain: timing not reported | No study reported on this outcome for both the intervention and comparator groups. | |||||
| Ear pain at the time of intratympanic injection | Comparison not applicablec | One study reported no participants with ear pain at the time of intratympanic injection (0/24) | 44 (one study) | Not calculable | ⊕⊝⊝⊝ very low6 | The evidence is very uncertain regarding the risk of ear pain at the time of intratympanic corticosteroid injection as secondary treatment. |
| *The basis for the assumed risk is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| aSix studies recruited participants suffering from sudden sensorineural hearing loss after treatment failure with systemic steroids. The incidence of improvement for the control group in these six studies was 6.96%. We have used 70 per 1000 to express the assumed risk. bOnly the most widely reported adverse events are described here. For adverse events that could feasibly occur in either group, we have only included the studies that provided a rate for both groups. For adverse events that could only occur in one group, we have only included the studies that reported the rate in that group, and presented these as a range. A full description of adverse event data is available for reference in Table 3. cComparisons between patients receiving intratympanic therapy and those receiving only systemic therapy were regarded as invalid for the following adverse events: persistent tympanic membrane perforation, vertigo observed at the time of intratympanic injection and ear pain observed at the time of intratympanic injection. This is explained in Data extraction and management. dThis includes participants who received placebo intratympanic injection.
GRADE Working Group grades of evidence | ||||||
| 1Downgraded one level due to risk of bias: we rated one study contributing moderate weight to the overall effect estimate as having high risk of bias due to incomplete outcome data. All studies were at high risk of other bias. Downgraded one level due to imprecision: the 95% CI for the effect overlaps the threshold for clinical relevance and the sample size is smaller than the optimal information size (taken as 400 participants). One study included treatment in the comparator arm with vitamin B, vasodilators and benzodiazepines (Ho 2004). However, as the weight of this study in the meta‐analysis was low and exclusion of the study made little difference to the effect estimate we did not downgrade for indirectness. 2Downgraded one level due to risk of bias: we rated two studies as being at high risk of bias due to selective reporting and one study was at high risk of bias for incomplete outcome data. All studies were at high risk of other bias. Downgraded one level due to imprecision: the total number of events is smaller than the optimal information size (taken as 300 events). 3Downgraded one level due to risk of bias: we rated one study contributing moderate weight to the overall effect estimate as high risk of bias due to incomplete outcome data, and one other study as high risk of bias because of selective reporting. All studies were at high risk of other bias. Downgraded one level due to imprecision: the 95% CI for the effect overlaps the threshold for clinical relevance and the sample size is smaller than the optimal information size (taken as 400 participants). 4Downgraded one level due to risk of bias: we rated one study as high risk of bias because of selective reporting; we rated one study as high risk of bias because of incomplete outcome data; we rated three studies as uncertain for random sequence generation, allocation concealment and blinding. Downgraded two levels because of imprecision: the number of events is smaller than the optimal information size (taken as 300 events) and an effect estimate could not be calculated. 5Downgraded one level due to risk of bias: we rated one study as high risk of bias because of selective reporting; we rated one study as high risk of bias because of incomplete outcome data; we rated two studies as uncertain for random sequence generation, allocation concealment and blinding. Downgraded two levels because of imprecision: the number of events is smaller than the optimal information size (taken as 300 events) and an effect estimate could not be calculated. 6Downgraded two levels because of imprecision: the number of events is smaller than the optimal information size (taken as 300 events) and an effect estimate could not be calculated. Downgraded one level due to indirectness: single study from a single setting, which may not adequately represent all patients with ISSNHL. | ||||||
| Adverse event reported | Study | How reported | Details of recovery | Rate in intervention group (%) | Rate in comparator group (%) | RR (95% CI) |
|---|---|---|---|---|---|---|
| TM perforation | No residual TM perforation | NA | 0/15 (0) | NA | NA | |
|
| Persistent TM perforation | No hearing loss in the affected ear. The perforation was treated successfully with a paper patch. | 1/24 (4.2) | NA | NA | |
|
| One patient (unclear which group) had a major catheter dislocation with perforation of ear drum. Note: both groups received IT injection, either corticosteroid or normal saline. | Small ear drum perforation was closed with a myringoplasty | NR | NR | NA | |
|
| Transient TM perforation | Healed spontaneously by follow‐up 1 month later | 1/27 (3.7) | NR | NA | |
|
| No TM perforation was noticed at last visit | NA | 0/19 (0) | NA | NA | |
| Vertigo/dizziness | Complained of vertigo immediately after injection | Recovered after 2 hours of rest | 1/15 (6.7) | NA | NA | |
|
| Three patients complained of vertigo [...] during the injections | Resolved within minutes | 3/24 (12.5) | NA | NA | |
|
|
| No disequilibrium | NA | 0/24 (0) | NR | NA |
|
|
| No dizziness for more than 24 hours | NA | 0/24 (0) | NR | NA |
|
| One patient (unclear which group) with increase in vertigo | Resolved | NR | NR | NA | |
|
| Temporary dizziness experienced by one‐third of subjects (unclear how many each group). Note: both groups received IT injection, either corticosteroid or normal saline. | Relieved by resting for a short time. Three participants quit the trial because of uncomfortable dizziness (unclear how many each group). | NR | NR | NA | |
| Tinnitus | Three patients complained of vertigo or an increase in tinnitus during the injections | Resolved within minutes | 3/24 (12.5) | NA | NA | |
| Hearing loss | The injection did not induce an increase in […] hearing loss […] for greater than 24 h | NA | 0/24 (0) | NR | NA | |
|
| No participant experienced a decrease in hearing of 10 dB or more | NA | 0/27 | 0/28 | NA | |
| Ear pain | The injection did not induce an increase in ear pain […] | NA | 0/24 (0) | NA | NA | |
|
| Two patients (unclear how many each group) with ear pain. Note: both groups received IT injection, either corticosteroid or normal saline. | Resolved | NR | NR | NA | |
|
| A mild ear pain occurring the first hour post‐injection | Easily controlled with common analgesics | NR | NA | NA | |
| Other | One of 15 patients had acne | NR | 1/15 (6.7) | NR | NA | |
|
| No serious complications such as chronic otitis media, disequilibrium or dysgeusia developed | NA | 0/24 (0) | NR | NA | |
|
| One patient (unclear which group) with each of: ear canal skin defect, steroid acne, nausea after antibiotic intake, gastroenteritis, hypokalaemia, pump battery failure and viral conjunctivitis. Three patients with headache (unclear how many in each group; one considered as 'possibly', 'probably' or 'very likely' related to the study) and 3 (unclear how many each group) with increased liver function tests (probably due to antibiotics). | Resolved | NR | NR | NA | |
|
| Severe nausea or vomiting was not experienced by any of the participants after the injection therapy | NA | 0/27 | 0/28 | NA | |
|
| No infection was noticed in any of the patients at their last visit | NA | 0/19 (0) | 0/18 (0) | NA | |
| TM: tympanic membrane; NA: not applicable; NR: not reported; IT: intratympanic | ||||||
| Intratympanic plus systemic corticosteroids (combined therapy) versus systemic corticosteroids alone as secondary therapy | ||||||
| Patient or population: sudden sensorineural hearing loss Settings: after treatment failure with systemic steroids Intervention: combination of intratympanic and systemic steroid therapy Comparison: systemic steroid therapy | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | No of participants | Relative effect | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Systemic therapy (assumed risk) | Combined therapy (corresponding risk) | |||||
| Change in hearing threshold determined by PTA | No studies reported this outcome. | |||||
| Proportion of patients whose hearing is improved | 205 per 1000a | 459 per 1000 (226 to 933) | 76 (1 study) | RR 2.24 (95% CI 1.10 to 4.55) | ⊕⊝⊝⊝ | Combined therapy may increase the proportion of patients whose hearing is improved compared to systemic corticosteroids alone (as secondary therapy), but the evidence is very uncertain. |
| Final hearing threshold determined by PTA | No studies reported this outcome. | |||||
| Adverse eventsb | Events in control group | Events in intervention group | No of participants | Relative effect | Certainty of the evidence | Comments |
| Persistent tympanic membrane perforation | Comparison not appropriatec | One study reported a rate of 8.1% (3/37) | 76 (1 study) | Not calculable | ⊕⊝⊝⊝ very low2 | The risk of tympanic membrane perforation among those who receive intratympanic corticosteroid combined with systemic corticosteroid as primary treatment is very uncertain. |
| Vertigo/dizziness: timing not reported | No studies reported this outcome. | |||||
| Vertigo/dizziness: at the time of injection | Comparison not appropriatec | No study reported a rate in the intervention group. | ||||
| Ear pain: timing not reported | No studies reported this outcome. | |||||
| Ear pain: at the time of injection | Comparison not appropriatec | No study reported a rate in the intervention group. | ||||
| *The basis for the assumed risk is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| aOne study recruited participants suffering from sudden sensorineural hearing loss after treatment failure with systemic steroids. The incidence of improvements was 20.51%. We have used 205 per 1000 to express the assumed risk. bOnly the most widely reported adverse events are described here. For adverse events that could feasibly occur in either group, we have only included the studies that provided a rate for both groups. For adverse events that could only occur in one group, we have only included the studies that reported the rate in that group, and presented these as a range. A full description of adverse event data is available for reference in Table 4. cComparisons between patients receiving intratympanic therapy and those receiving only systemic therapy were regarded as invalid for the following adverse events: persistent tympanic membrane perforation, vertigo observed at the time of intratympanic injection and ear pain observed at the time of intratympanic injection. This is explained in Data extraction and management.
GRADE Working Group grades of evidence | ||||||
| 1Downgraded two levels due to risk of bias: we judged the study to be at high risk of selection bias, performance bias, incomplete outcome data, selective reporting and other bias. Downgraded two levels due to imprecision: the 95% CI overlaps the threshold for clinical relevance and the total number of events is smaller than the optimal information size (taken as 300 events). 2Downgraded two levels due to risk of bias: we judged the study to be at high risk of bias because of selection bias, concern about blinding, incomplete outcome data and selective reporting. Downgraded two levels due to imprecision: the number of events is smaller than the optimal information size (taken as 300 events) and an effect estimate could not be calculated. Downgraded one level because of indirectness: provision of data by only a single study from a single setting, which may not adequately represent all patients with ISSNHL. | ||||||
| Adverse event reported | Study | How reported | Details of recovery | Rate in intervention group (%) | Rate in comparator group (%) | RR (95% CI) |
|---|---|---|---|---|---|---|
| TM perforation | Three patients had small eardrum perforations | Successful closure by simple treatment | 3/37 (8.1) | NA | NA | |
| Vertigo | Second frequent complaint: transient vertigo after the drug had been injected into the ear | Not a severe problem if the drug was heated in 37°C water before injection and the vertigo disappeared after a few minutes or under 30 minutes | NR | NA | NA | |
| Ear pain | Most frequent complaint | Easily controlled by the oral administration of paracetamol 30 minutes before the local infusion of the methylprednisolone | NR | NA | NA | |
| Hearing loss* | No loss in hearing related to the treatment (in either group) | NA | 0/37 (0) | 0/39 (0) | NA | |
| Other | One patient had tongue paresthesia (unclear which group) | Resolved after 2 weeks | NR | NR | NA | |
|
|
| No infections were observed (unclear which group) | NA | NR | NR | NA |
|
|
| Long‐term complications did not occur in any patients who received the transtympanic injections | NA | 0/37 (0) | NR | NA |
| *Hearing loss defined as ≥ 15 dB worsening in pure tone audiometry or ≥ 15% worsening of speech discrimination score. NA: not applicable; NR: not reported; TM: tympanic membrane | ||||||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1.1 Mean change in pure tone average (PTA) Show forest plot | 10 | 701 | Mean Difference (IV, Fixed, 95% CI) | ‐5.93 [‐7.61, ‐4.26] |
| 1.2 Proportion improved Show forest plot | 14 | 972 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.04 [0.97, 1.12] |
| 1.3 Final PTA Show forest plot | 7 | 516 | Mean Difference (IV, Fixed, 95% CI) | ‐3.31 [‐6.16, ‐0.47] |
| 1.4 Change in hearing threshold with speech audiometry Show forest plot | 1 | 98 | Mean Difference (IV, Fixed, 95% CI) | ‐8.85 [‐19.58, 1.88] |
| 1.5 Speech audiometry: additional outcomes Show forest plot | 3 | Mean Difference (IV, Fixed, 95% CI) | Subtotals only | |
| 1.5.1 Speech discrimination score: change from baseline | 1 | 98 | Mean Difference (IV, Fixed, 95% CI) | 15.64 [1.57, 29.71] |
| 1.5.2 Speech discrimination score: endpoint | 1 | 35 | Mean Difference (IV, Fixed, 95% CI) | 6.00 [‐20.88, 32.88] |
| 1.5.3 Word recognition score: change from baseline | 1 | 250 | Mean Difference (IV, Fixed, 95% CI) | ‐0.60 [‐9.28, 8.08] |
| 1.6 Frequency‐specific changes in PTA Show forest plot | 6 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 1.6.1 0.25 kHz | 2 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 1.6.2 0.5 kHz | 3 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 1.6.3 1 kHz | 3 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 1.6.4 2 kHz | 3 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 1.6.5 3 kHz | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 1.6.6 4 kHz | 3 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 1.6.7 8 kHz | 2 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 1.6.8 Frequency range: low | 3 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 1.6.9 Frequency range: mid | 3 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 1.6.10 Frequency range: high | 2 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 1.7 Adverse events Show forest plot | 2 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 1.7.1 Ear pain | 2 | 289 | Risk Ratio (M‐H, Fixed, 95% CI) | 15.68 [6.22, 39.49] |
| 1.7.2 Otitis media | 1 | 250 | Risk Ratio (M‐H, Fixed, 95% CI) | 3.28 [0.70, 15.49] |
| 1.7.3 Vertigo/dizziness | 1 | 250 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.53 [1.41, 4.54] |
| 1.7.4 Blood glucose problems | 2 | 289 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.54 [0.35, 0.85] |
| 1.7.5 Mood change | 2 | 289 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.22 [0.13, 0.37] |
| 1.7.6 Sleep change | 2 | 289 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.19 [0.10, 0.36] |
| 1.7.7 Appetite change | 2 | 289 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.20 [0.09, 0.44] |
| 1.7.8 Weight change | 2 | 289 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.28 [0.13, 0.61] |
| 1.7.9 Dry mouth | 2 | 289 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.15 [0.06, 0.35] |
| 1.7.10 Any adverse event | 1 | 250 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.03 [0.94, 1.12] |
| 1.7.11 Serious adverse event | 1 | 250 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.13 [0.35, 3.59] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 2.1 Mean change in pure tone average (PTA) Show forest plot | 6 | 435 | Mean Difference (IV, Fixed, 95% CI) | ‐8.55 [‐12.48, ‐4.61] |
| 2.2 Proportion improved Show forest plot | 10 | 788 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.27 [1.15, 1.41] |
| 2.3 Final PTA Show forest plot | 3 | 194 | Mean Difference (IV, Fixed, 95% CI) | ‐9.11 [‐16.56, ‐1.67] |
| 2.4 Change in hearing threshold with speech audiometry Show forest plot | 1 | 98 | Mean Difference (IV, Fixed, 95% CI) | ‐7.59 [‐20.22, 5.04] |
| 2.5 Speech audiometry: additional outcomes Show forest plot | 3 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 2.5.1 Speech discrimination score: change from baseline | 2 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 2.5.2 Speech discrimination score: endpoint | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 2.6 Frequency‐specific changes with PTA Show forest plot | 4 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 2.6.1 0.25 kHz | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 2.6.2 0.5 kHz | 2 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 2.6.3 1 kHz | 2 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 2.6.4 2 kHz | 2 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 2.6.5 3 kHz | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 2.6.6 4 kHz | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 2.6.7 8 kHz | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 2.6.8 Frequency range: low | 2 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 2.6.9 Frequency range: mid | 2 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 2.6.10 Frequency range: high | 2 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 3.1 Mean change in PTA Show forest plot | 7 | 280 | Mean Difference (IV, Fixed, 95% CI) | ‐9.07 [‐11.47, ‐6.66] |
| 3.2 Proportion improved Show forest plot | 6 | 232 | Risk Ratio (M‐H, Fixed, 95% CI) | 5.55 [2.89, 10.68] |
| 3.3 Final PTA Show forest plot | 5 | 203 | Mean Difference (IV, Fixed, 95% CI) | ‐11.09 [‐17.46, ‐4.72] |
| 3.4 Change in hearing threshold with speech audiometry Show forest plot | 1 | 21 | Mean Difference (IV, Fixed, 95% CI) | ‐12.80 [‐30.17, 4.57] |
| 3.5 Speech audiometry: additional outcomes Show forest plot | 1 | 21 | Mean Difference (IV, Fixed, 95% CI) | 19.90 [0.41, 39.39] |
| 3.5.1 Maximum speech discrimination: change from baseline | 1 | 21 | Mean Difference (IV, Fixed, 95% CI) | 19.90 [0.41, 39.39] |
| 3.6 Frequency‐specific changes with PTA Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 3.6.1 0.25 kHz | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 3.6.2 0.5 kHz | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 3.6.3 1 kHz | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 3.6.4 2 kHz | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 3.6.5 3 kHz | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 3.6.6 4 kHz | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 3.6.7 6 kHz | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 3.6.8 8 kHz | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 4.1 Proportion improved Show forest plot | 1 | 76 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.24 [1.10, 4.55] |
| 4.2 Speech audiometry: additional outcomes Show forest plot | 1 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 4.2.1 Improvement in speech discrimination score | 1 | 76 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.41 [1.12, 5.18] |
