Types of studies

We will include randomised controlled trials and quasi‐randomised studies, where the unit of analysis is the individual participant. We will use the first treatment period of cross‐over trials and treat this as a parallel‐arm trial, providing the study reporting permits.

Types of participants

Adults (≥ 18 years old) with mild or moderate hearing loss, as defined by pure‐tone thresholds in the better‐hearing ear averaged across four frequencies (0.5 kHz, 1.0 kHz, 2.0 kHz and 4.0 kHz) of 26 to 40 dB HL (mild hearing loss) and 41 to 60 dB HL (moderate hearing loss). In the absence of confirmation that all participants in a study meet these criteria (i.e. where participant‐level data are not reported or cannot be obtained), we will include studies where the reported participant characteristics for the mean four‐frequency average fall within the range for either mild or moderate hearing loss, as described above. If a mean frequency average is offered for a combination of frequencies other than 0.5 kHz, 1.0 kHz, 2.0 kHz and 4.0 kHz, we will use studies where the reported value falls between 26 to 40 dB HL (mild hearing loss) and 41 to 60 dB HL (moderate hearing loss). If only qualitative descriptions of mild and moderate hearing loss are given with no supporting audiometric data, we will include such studies but will not include them in the meta‐analysis.

Types of interventions

Acoustic hearing aids, irrespective of where they are worn or the type of technology (analogue or digital).

We will exclude hearing aids or implantable devices whose primary purpose is to deliver bone conduction sound or those that detect and deliver sound via air conduction to the contralateral ear.

The comparisons of interest are hearing aids versus either a passive control (no intervention, waiting list control; we will pool these in meta‐analysis) or an active control that involves:

• information/education only, listening tactics and communication training (we will pool these in meta‐analysis);

• assistive listening devices; or

• auditory training (we will analyse these second two in separate meta‐analyses).

We will not consider studies where the intervention is delivered in a group setting.

Types of outcome measures

We will analyse these outcomes in the review, but we will not use them as a basis for including or excluding studies. We will analyse the data at the trial endpoint, with a subgroup analysis to compare different trial endpoints.

Electronic searches

Published, unpublished and ongoing studies will be identified by searching the following databases from their inception:

• Cochrane Register of Studies (CRS) ENT Trials Register (search to date);

• Cochrane Central Register of Controlled Trials (CENTRAL, current issue);

• Ovid MEDLINE (1946 to date);

• Ovid MEDLINE (In‐Process & Other Non‐Indexed Citations);

• PubMed (as a top up to searches in Ovid MEDLINE);

• Ovid EMBASE (1974 to date);

• EBSCO CINAHL (1982 to date);

• Ovid CAB abstracts (1910 to date);

• LILACS (search to date);

• KoreaMed (search to date);

• IndMed (search to date);

• PakMediNet (search to date);

• Web of Knowledge, Web of Science (1945 to date);

• CNKI (searched via Google Scholar to date);

• ClinicalTrials.gov (www.clinicaltrials.gov) (search via the CRS to date);

• ICTRP (search to date);

• ISRCTN (www.isrctn.com) (search to date);

• Google Scholar (search to date);

• Google (search to date).

The subject strategies for databases will be modelled on the search strategy designed for CENTRAL (Appendix 1). Where appropriate, these will be 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)).

Searching other resources

We will scan the reference lists of identified publications for additional trials and contact trial authors if necessary. In addition, the Trials Search Co‐ordinator will search PubMed, TRIPdatabase, The Cochrane Library and Google to retrieve existing systematic reviews relevant to this systematic review, so that we can scan their reference lists for additional trials. We will search for conference abstracts using the Cochrane ENT Trials Register and EMBASE.

Selection of studies

Material downloaded from electronic sources will include details of author, institution or journal of publication and abstract. MF, DH and Fiona Barker (FB) will independently screen each study against the criteria for including studies to determine their eligibility for inclusion in the review. Where the decision about any one study is not unanimous among the screening authors, we will acquire the full article for further inspection. Once the full articles are obtained, we will decide whether the studies meet the review criteria. We will make a final decision by consensus.

Data extraction and management

MF and FB will independently extract data from the articles. We will record the extracted data on a review‐specific form that we will develop and assess for suitability through pilot testing prior to independent data extraction. Where data to be extracted are also described in the text of an article, whether in the main body or in a table, PK will resolve discrepancies between the two independent extractions. Where data to be extracted are also described graphically in an article, we will use the average of the two independent extractions provided there is agreement regarding their derivation (see also Dealing with missing data).

Information to be extracted will include: trial design, setting, methods of randomisation and blinding, power, inclusion and exclusion criteria, type of intervention and control, time since hearing aid fitting, and duration of follow‐up, outcome measures and statistical tests.

For both the intervention and control groups, data extraction will include: baseline characteristics of participants (number, sex, age and details of hearing loss (mean, standard deviation, range), and details of any attrition or exclusion. For the intervention group we will extract details of hearing aids (where worn, technology type, manufacturer, unilateral or bilateral hearing aid fittings).

Outcome measure data to be extracted will include: group means and standard deviations at pre‐ and post‐intervention and follow‐up, number of participants and results of statistical tests of between‐group comparisons.

We will contact the authors of a study should further information be required, which cannot be extracted from the published report.

After independent data extraction by MF and FB, MEJ will review the extracted data for disagreements, and revisit and discuss the relevant studies with MF and FB until a consensus is reached.

Assessment of risk of bias in included studies

MF and PK will undertake assessment of the risk of bias of the included trials independently, with the following taken into consideration, as guided by theCochrane Handbook for Systematic Reviews of Interventions (Handbook 2011):

• sequence generation;

• allocation concealment;

• blinding of (i) participants and study personnel (performance bias), and (ii) outcome assessment (detection bias);

• incomplete outcome data;

• selective outcome reporting; and

• other sources of bias.

We will use the Cochrane 'Risk of bias' tool in RevMan 5.3 (RevMan 2014), which involves describing each of these domains as reported in the trial and then assigning a judgement about the risk of bias on the outcomes measured as a result of each entry: 'low', 'high' or 'unclear' risk of bias.

Measures of treatment effect

We will express the size of the difference in treatment effect between a treatment group and a control group at a trial endpoint in terms of the standardised mean difference (SMD). The calculation of the effect size will use the pooled standard deviation. We will also report the 95% confidence interval (CI) for each SMD. An effect size greater than 0 indicates that a larger treatment effect was observed in the treatment group relative to the control group. For binary data, including adverse effects, we will preferably express the treatment effects as risk ratios (RR).

Unit of analysis issues

The unit of analysis will be the participant. We do not anticipate that participant‐level data will generally be available, but suitable summary statistics should be provided. As stated above, in the case of cross‐over trials we will only include the first period.

Dealing with missing data

We will make efforts to contact the corresponding author of an included study to obtain any missing data. Where data can only be estimated by reading off plots, we will achieve this as detailed in the Data extraction and management section. If standard deviations are not reported or provided by the authors we will estimate standard deviations in RevMan 5.3 using available data, such as standard errors, 95% CIs, P values and t values. If data are not available for subgroups of interest then we will request these data from study authors; likewise if data, for example, standard deviations (or error bars on plots) are not available. Where missing data cannot be obtained, we will describe the methodology used to account for these missing data and investigate the mechanism by which data might be missing (e.g. whether missing completely at random). Where possible, we will perform analysis on an available case analysis basis, unless this is not feasible given the use of imputation in individual studies. If possible, we will report the extent of the missing data within studies.

Assessment of heterogeneity

We will assess heterogeneity among treatment effect sizes using RevMan 5.3 (RevMan 2014), and we will express this in terms of the I² statistic. We will assess statistical significance using a Chi² test with K‐1 degrees of freedom. We will quantify heterogeneity in terms of the I² statistic with low, medium and high ranges of 0% to 40%, 41% to 60% and 61% to 100%, respectively.

Assessment of reporting biases

We will assess publication bias by examining a funnel plot in which the size of treatment effects is plotted against their variability. We will quantify deviation from the expected symmetrical pattern by calculating the number of studies required to achieve symmetry using the 'trim and fill' method (Duval 2000).

Data synthesis

We will conduct random‐effects meta‐analyses of the SMDs using RevMan 5.3 (RevMan 2014), as we anticipate significant heterogeneity across treatment effects between studies. For each meta‐analysis, we will report a summary effect size estimate in terms of the SMD (and where appropriate risk ratio) together with its 95% CI. We will calculate summary effects using the generic inverse variance procedure.

Subgroup analysis and investigation of heterogeneity

If heterogeneity is identified we will use subgroup analyses to assess possible sources. These factors will include age at hearing aid fitting, sex and degree of hearing loss (i.e. mild or moderate). Degree of hearing loss will be based on the better ear hearing thresholds averaged across 0.5 kHz, 1.0 kHz, 2.0 kHz and 4.0 kHz and we will classify this as mild and moderate for hearing thresholds between 26 to 40 dB hearing level (HL) and 41 to 60 dB HL, respectively. Where studies do not report separately by subgroup, or do not fall completely within one particular subgroup, we will allocate studies to a subgroup where there is evidence that a majority of participants fall within that group. Age will be defined as older adults (> 55 years) and younger adults (<= 55 years). If a study includes only one sex, we propose that those studies go entirely into one subgroup, and if a study reports males and females separately they can contribute to both subgroups of an analysis. If, however, a study only reports a mixed group, then we will remove this from the analysis. Time between fitting and trial endpoint may also represent a source of heterogeneity and so a subgroup analysis will also compare trials with endpoints at up to three months, over three months to six months and six months or more.

Sensitivity analysis

We will perform sensitivity analyses to informally test the robustness of assumptions from the data.

We will use sensitivity analyses where there is any uncertainty regarding any aspect of the included studies in terms of randomisation (random/quasi‐random), missing data (greater than 30% at the primary endpoint) and description of hearing loss (mild/moderate).