Albu 2016.

Study characteristics
Methods	Parallel‐group, triple‐blinded RCT with 24 months duration of treatment and follow‐up
Participants	Setting:  Multicentre trial conducted at 4 sites in Romania and 1 site in Italy. Participants recruited from ENT or otology clinics and managed as outpatients. Trial conducted from January 2009 until June 2013.  Sample size: Number randomised: 66 participants Number completed: 62 participants Participant baseline characteristics Age:  Not reported Gender:  High‐dose betahistine group: 15 males (45.5%); 18 females (54.5%) Control group: 12 males (36.4%); 21 females (63.6%) Probable/definite Ménière's disease:  All participants had definite disease Duration of disease:  Not reported Attack frequency at baseline:  Not reported. All participants were required to have an attack frequency of at least 4 episodes per month during the 3 preceding months.  Hearing loss at baseline:  Betahistine group: mean hearing loss 54.6 dB HL (SD 15.2) Control group: mean hearing loss 51.4 (SD 13.6) Measure of tinnitus at baseline:  Betahistine group: mean THI 28.3 (SD 14.8) Control group: mean THI 27.7 (SD 16.7) Number of participants with bilateral disease: All participants had unilateral disease Inclusion criteria: Adult participants with unilateral, definite Ménière's disease according to the AAO‐HNS 1995 criteria. Episodes of spontaneous vertigo lasting between 20 minutes and 12 hours. Mean of 4 or more vertigo spells per month during the 3 months before trial entry. Failure to control symptoms with a 6‐month trial of low‐salt diet, caffeine and nicotine avoidance. Exclusion criteria: Bilateral disease. Other peripheral or central vestibular syndromes. Middle ear pathology. Noise‐induced hearing loss. Previous IT gentamicin or corticosteroid use. Previous ablative ear surgery. Allergy to betahistine. Diagnosis of Ménière's disease: AAO‐HNS 1995 criteria for definite Ménière's disease
Interventions	Intervention (n = 33 randomised, n = 30 completed) High‐dose betahistine; participants received 48 mg betahistine 3 times daily (total 144 mg) Comparator (n = 33 randomised, n = 32 completed) Placebo; participants received identical appearing placebo pills Duration of treatment was not provided but was assumed to be for the duration of the trial (2 years) Background interventions administered to all participants All participants received intratympanic dexamethasone: "Dexamethasone was injected under the microscope [...]: in the supine position, the patient turned the head 45° toward the unaffected ear. Local anesthesia of the tympanic membrane was achieved and dexamethasone (4 mg/mL [total quantity not stated]) was injected through a 22‐gauge spinal needle and 1‐mL syringe to fill the middle ear. Patients were instructed to keep the supine position with the treated ear facing upward for 30 min avoiding swallowing or talking. The ITD protocol consisted of three consecutive daily injections."
Outcomes	Primary outcomes relevant to this review: Improvement in vertigo Assessed using the AAO‐HNS 1995 class of vertigo  Change in vertigo Not reported  Serious adverse events Not apparently systematically assessed or reported. Narrative summary of adverse events only.  Secondary outcomes relevant to this review: Disease‐specific health‐related quality of life Assessed with the  AAO‐HNS 1995 FLS Hearing Assessed with pure tone audiometry at 24 months  Tinnitus  Assessed with the THI at 24 months  Other adverse effects Not apparently systematically assessed or reported. Narrative summary of adverse events only.  Other outcomes reported in the study: Speech discrimination score Survival curves for attainment of complete and substantial vertigo control Hearing assessed as improvement/unchanged/worsened
Funding sources	Quote: "No funding was received for this study"
Declarations of interest	There is no declaration of interest
Notes	Research integrity checklist: No retractions or expressions of concern were identified No prospective trial registration was identified; the authors are unable to supply us with a copy of the original trial protocol  Baseline characteristics of the groups are not excessively similar  There is little loss to follow‐up, despite the length of the trial (2 years); the authors state that this is because participants were receiving free treatment during the trial  The study is free from implausible results  Identical numbers of participants were randomised to each group, but block randomisation is not described
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "For random assignment of patients, a computer‐generated list of random numbers was employed. Randomization was performed by an only investigator (LM) one day before the injection." Comment: computer‐generated randomisation list.
Allocation concealment (selection bias)	Unclear risk	Comment: no methods to conceal allocation were reported.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "Group A received a combination of ITD and identical‐appearing placebo pills". "Both the surgeons and the patients were blinded to the treatment".  Comment: likely that blinding of participants and personnel was adequate if placebo pills were identical.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Quote: "Audiometric testing and completion of questionnaires were performed by researchers blinded to the surgeons. Throughout the study, all patients kept a diary recording monthly the occurrence of every vertiginous attack"  Comment: outcomes were assessed by (blinded) participants and blinded researchers. However, we note that side effects that should have been likely to occur in both groups over the 2‐year follow‐up (nausea, headache, diarrhoea) were only reported for the betahistine group. This may be due to selective reporting, but there is a concern that investigators recorded these symptoms because they were aware of the group allocation. There is no description of how these adverse events were collected and whether this was identical in both groups.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Quote: "Three (sic) patients (1 in Group A and 3 in Group B) were lost to follow‐up."  Comment: there was only a moderate difference in the numbers lost to follow‐up across groups (9% in the betahistine group, and 3% in the placebo group). Given the large difference in effect size for the primary outcome (complete vertigo control achieved in 14 patients (44%) from Group A and in 22 patients (73.3%) from Group B, statistically significant (P = 0.01)), it is unlikely that bias would arise from incomplete outcome data.
Selective reporting (reporting bias)	Unclear risk	Comment: no published protocol was identified and the authors are unable to supply a copy of the original trial protocol.
Other bias	High risk	Comment: repeated intratympanic steroid injections were permitted if vertigo was not substantially/completely controlled. Additional treatments were also permitted through the study. It is unclear whether these treatments were offered equally and consistently to the 2 groups of participants. For example, the study states that "If complete or substantial vertigo control was not accomplished, another sequence of [IT dexamethasone] was offered. In patients with persistent vertigo, despite repeated ITD injections, IT gentamicin injection or ablative surgery was offered".  5 participants in the placebo arm received 1 repeat injection, and 6 participants received 2 repeated injections. According to the methods it appears that those who did not achieve complete/substantial control (11 patients, at the end of the study) should have gone on to receive further IT injections, but this does not appear to have been performed. However, this may be due to follow‐up being insufficient.