Types of studies

We sought randomised controlled trials (RCTs) with a parallel or cross‐over design, and cluster‐RCTs. We included studies reported in full text, those published as an abstract only, and unpublished data. We included studies regardless of the date or language of publication.

Types of participants

We considered studies with men, aged 18 years or older, with obstructive sleep apnoea syndrome (OSAS) and erectile dysfunction (ED).

Types of interventions

We investigated the following interventions and comparators.

Types of outcome measures

Electronic searches

We identified studies from searches of the following databases and trial registries:

• Cochrane Airways Trials Register (Cochrane Airways 2021), via the Cochrane Register of Studies (searched 14 June 2021);

• Cochrane Central Register of Controlled Trials (CENTRAL), via the Cochrane Register of Studies (searched 14 June 2021);

• MEDLINE Ovid SP (1946 to 14 June 2021);

• Embase Ovid SP (1974 to 14 June 2021);

• LILACS (Literatura Latino‐Americana em Ciências de Saúde; 1982 to 14 June 2021);

• US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (www.clinicaltrials.gov; searched 14 June 2021);

• World Health Organization International Clinical Trials Registry Platform (apps.who.int/trialsearch; searched 14 June 2021).

We searched all sources from inception, with no restriction on language of publication. The database search strategies are listed in Appendix 2. The Cochrane Airways Information Specialist developed the search strategy in collaboration with the review authors.

Searching other resources

We checked the reference lists of all primary studies included and of review articles (that were found during the selection process) for additional references. We searched relevant manufacturers' websites for study information. We also searched specific conference proceedings for the British Association of Urological Surgeons (BAUS); European Association of Urology (EAU); and American Urological Association (AUA), searched 14 June 2021.

We searched for errata or retractions from included studies published in full text on PubMed on 14 June 2021.

Selection of studies

Two review authors (FTB, LESF, or MPS) independently screened the titles and abstracts of the search results, and coded them as 'retrieve' (eligible or potentially eligible, or unclear), or 'do not retrieve'. We retrieved the full‐text study reports of all potentially eligible studies, and two review authors (FTB, LESF or MPS) independently screened them for inclusion, recording the reasons for exclusion of ineligible studies. We resolved any disagreement through discussion, or if was required, we consulted a third review author (RR). We identified and excluded duplicates, and collated multiple reports of the same study, so that each study, rather than each report, was the unit of interest in the review. We recorded the selection process in sufficient detail to complete a PRISMA flow diagram and 'Characteristics of excluded studies' table (Moher 2009; Figure 1).

Figure 1

---

---

PRISMA study flow diagram

Data extraction and management

We used a standard data collection form for study characteristics and outcome data, which had been piloted on at least one study in the review, and contained the following characteristics from included studies.

• Methods: study design, total duration of study, details of any 'run‐in' period, number of study centres and location, study setting, withdrawals, and date of study

• Participants: N, mean age, age range, severity of condition, diagnostic criteria, comparison, concomitant medications, and excluded medications

• Outcomes: primary and secondary outcomes specified and collected, and time points reported

• Notes: funding for studies, and notable conflicts of interest of trial authors

Two review authors (FTB, DVP) independently extracted outcome data. We noted in the 'Characteristics of included studies' table if outcome data were not reported in a usable way. We resolved disagreements by consensus, or by involving a third review author (RR). One review author (FTB) transferred data into the Review Manager 5 file (RevMan Web 2020). A second review author (RR) spot‐checked study characteristics for accuracy against the study report.

If necessary (data were lacking), we contacted the authors of included trials by email to request further information. We presented the results of these communications in Appendix 3.

Assessment of risk of bias in included studies

Two review authors (FTB, DVP) independently assessed risk of bias for each study using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2020). We resolved any disagreements by discussion, or by involving another review author (RR). We assessed the risk of bias according to the following domains:

• Random sequence generation;

• Allocation concealment;

• Blinding of participants and personnel;

• Blinding of outcome assessment;

• Incomplete outcome data;

• Selective outcome reporting;

• Other bias.

We judged each potential source of bias as high, low, or unclear, and provided a quote from the study report, together with a justification for our judgement in the risk of bias table. We summarised the risk of bias judgements across different studies for each of the domains listed. We considered blinding and incomplete outcome data separately for different outcomes. Where information on risk of bias was related to unpublished data or correspondence with a trialist, we noted this in the risk of bias tables.

When considering treatment effects, we took into account the risk of bias for the studies that contributed to that outcome.

Measures of treatment effect

We analysed dichotomous data as risk ratio (RR) when the number of men experiencing a specified outcome, or when health‐related and sex‐related quality of life outcomes were presented as dichotomous data (using a cut‐off adopted by the trial authors, and based on the literature). For continuous data (health‐related and sex‐related quality of life), we used the mean difference (MD), or standardised mean difference (SMD, when different scales were used for the same outcome across the pooled studies). If data from rating scales were combined in a meta‐analysis, we ensured they were entered with a consistent direction of effect (e.g. lower scores always indicated improvement).

We undertook meta‐analyses only when this was meaningful; that was, if the treatments, participants, and the underlying clinical question were similar enough for pooling to make sense.

We described skewed data narratively (for example, as medians and interquartile ranges for each group).

When multiple trial arms were reported in a single study, we included only the relevant arms. When two comparisons (e.g. CPAP regimen A versus placebo, and CPAP regimen B versus placebo) were combined in the same meta‐analysis, we either combined the active arms, or halved the control group to avoid double‐counting.

If adjusted analyses were available (ANOVA or ANCOVA), we used these as a preference in our meta‐analyses.

In the case of cross‐over trials, we planned to use only results from the first treatment phase to avoid any carry‐over effect. For cluster‐RCTs, we planned to calculate the design effect using the intra cluster correlation coefficient (ICC).

We used intention‐to‐treat (ITT) or 'full analysis set' analyses when they were reported (i.e. when data were imputed for participants who were randomly assigned but did not complete the study), instead of complete or 'per protocol' analyses.

Unit of analysis issues

For dichotomous outcomes, we used participants, rather than events, as the unit of analysis (i.e. number of participants presenting with at least one serious adverse events, rather than the number of serious adverse events). However, if rate ratios were reported in a study, we analysed them on this basis. We only planned to meta‐analyse data from cluster‐RCTs if the available data were adjusted (or could be adjusted), to account for the clustering.

Dealing with missing data

We contacted investigators or study sponsors to verify key study characteristics and obtain missing numerical outcome data, when possible (e.g. when a study was identified as an abstract only). When this was not possible, and the missing data were thought to introduce serious bias, we took this into consideration in the GRADE rating for affected outcomes, and in our judgement of risk of bias.

Assessment of heterogeneity

We investigated clinical, methodological, and statistical heterogeneity among included trials. We did not pool effect estimates in a meta‐analysis if we identified substantial clinical or methodological heterogeneity among included studies. We identified statistical heterogeneity by visual inspection of the forest plots, and by using a standard Chi² test (significance level of 0.1). We quantified this heterogeneity by using the I² statistic; an I² statistic of 50% or more indicated substantial heterogeneity (Higgins 2020). When we found substantial heterogeneity, we attempted to determine possible reasons for it by individually exploring the trials included in the meta‐analysis, and by performing subgroup analyses.

Assessment of reporting biases

We planned to create funnel plots to explore possible publication biases for meta‐analysis with 10 or more studies. We did not investigate publication bias because of the low number of RCTs included in the meta‐analysis.

Data synthesis

We used a random‐effects model for meta‐analysis, and performed a sensitivity analysis with a fixed‐effect model. When we considered meta‐analysis to be inappropriate, we provided a narrative description of the study results.

Subgroup analysis and investigation of heterogeneity

We planned the following subgroup analyses for the primary outcomes.

• Partial versus complete adherence. We defined complete adherence as a mean of four or more hours of non‐invasive positive airway pressure device use per night, for at least one month, at any pressure level and, with any brand (Antic 2011; Sawyer 2011; Weaver 2007; Zimmerman 2006). We expected better results for men with complete adherence.

• Obese versus non‐obese participants. We defined obesity as body mass index (BMI) greater than 30. We expected better results for non‐obese participants.

• Older participants (60 years or older) versus younger participants (younger than 60 years). We expected better results for younger participants.

• Differences between CPAP, BIPAP, VPAP, or similar devices.

We planned to use the formal test for subgroup interactions available in Review Manager 5 (RevMan Web 2020).

We did not conduct any subgroup analyses because of the scarcity of studies included in each meta‐analysis.

Sensitivity analysis

We planned to carry out sensitivity analyses by removing trials with high risk of bias (those classified as having high risk of bias in at least one of the following domains: random sequence generation, allocation concealment, and blinding) from the primary outcome analyses.

We planned to test the robustness of results by repeating the analyses using a fixed‐effect model.

Summary of findings and assessment of the certainty of the evidence

We created a summary of findings table, using all primary and secondary outcomes. We used the five GRADE considerations (risk of bias, consistency of effect, imprecision, indirectness, and publication bias) to assess the certainty of the body of evidence as it related to the studies that contributed data for each prespecified outcome. We used the methods and recommendations described in Section 8.5 and Chapter 12 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2020), and GRADEpro GDT software (GRADEpro GDT). We used footnotes to justify all decisions to downgrade our assessments of the certainty of evidence, and we made comments to aid the reader's understanding of the review where necessary.

Two review authors (FTB, RR) independently rated the certainty of evidence for each outcome as high, moderate, low, or very low; discrepancies were resolved by a third author (LESF).