Melin 1987.

Methods	2‐arm, non‐blinded, multi‐centre, parallel‐group RCT, with a 6 weeks follow‐up
Participants	Location: Sweden, 1 site Setting of recruitment and treatment: hearing clinic at a Swedish university hospital Sample size: Number randomised: 39 (20 intervention, 19 control) Number completed: 39 (attrition 0%) Participant (baseline) characteristics: Age: mean 72.7 years (SD 10.6) Gender: 13 male, 26 female Main diagnosis: hearing loss (PTA averaged across 0.5, 1.0, 2.0, 4.0 kHz) left = 40.5 dB HL (SD 11.9) right = 39.05 (SD 13.1) Other important effect modifiers: none Inclusion criteria: hearing loss to a degree that hearing aids were needed, no prior hearing aid experience, tinnitus duration for more than 6 months Exclusion criteria: none noted
Interventions	Intervention group (n = 20): hearing aids (manufacturer: Widex, Rexton, Oticon, Siemens, Philips, Danavox), unilateral fits (95%) 6 weeks post‐baseline. Fitting not specified. Comparator group (n = 19): waiting list controls, no hearing aids up to 6 weeks post‐baseline Use of additional interventions (common to both treatment arms): none up to 6 weeks post‐baseline, then both groups had hearing aids
Outcomes	Primary outcomes: Hearing Scaling (easy, fairly easy, fairly difficult, difficult), 6 weeks post‐fitting Secondary outcomes: none specified Reported outcomes: as for primary outcomes Hearing Scaling assessment was not included in the analysis of listening abilities because not all of the data required were available in the paper and were no longer available.
Funding sources	Bank of Sweden Tercentenary Foundation (grant No. 83/16) and grants from Stifrelsen, Stockholm and Oticon Foundation, Copenhagen
Declarations of interest	None noted
Notes	6 weeks after baseline, hearing aids were fitted to the control group and the study continued to 10 weeks follow‐up for each group. Hearing aids were provided at no cost.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Insufficient information was provided in the paper. Quote: "To prevent bias, the random allocations of the subjects were done after their first interview according to a randomisation plan". However, contact with the authors revealed randomisation was most likely done in blocks of 10 (5 participants experimental group, 5 participants control group) to recruit groups of the same size.
Allocation concealment (selection bias)	Low risk	Insufficient information was provided in the paper. Quote: "To prevent bias, the random allocations of the subjects were done after their first interview according to a randomisation plan". However, contact with the authors revealed that the allocation to group was concealed by using pre‐prepared opaque envelopes.
Blinding of participants and personnel (performance bias) All outcomes	High risk	No blinding, therefore patients and personnel were aware of the intervention. Possible that the hearing aid group were treated differently compared to the waiting list group in other aspects.
Blinding of outcome assessment (detection bias) All outcomes	High risk	No blinding and outcomes were likely to have been influenced by the lack of blinding.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	No participant dropouts, but insufficient reporting of whether all data points from the hearing scale were completed by group. Historic records not available.
Selective reporting (reporting bias)	Low risk	Study protocol not available but the authors indicated that no measures other than the listening scaling task were included in the study.
Other bias	Unclear risk	Waiting list controls have a risk of overestimating the benefit of an intervention. However, all participants in the waiting lists were to receive the active intervention at the end of the experimental phase (not conditional on severity at the end of the study).