Types of studies

Published and unpublished randomised controlled trials (RCT) were eligible for inclusion. We excluded non‐randomised studies and quasi‐randomised trials.

Types of participants

1. Subfertile women, with otherwise unexplained infertility, in whom routine imaging did not show uterine cavity abnormalities, who wished to conceive spontaneously

2. Subfertile women, in whom routine imaging did not show uterine cavity abnormalities, and before treatment with IUI.

3. Women in whom routine imaging did not show uterine cavity abnormalities, and before treatment with IVF.

We excluded subfertile women with suspected uterine cavity abnormalities (present on any imaging techniques), as this topic is covered by another Cochrane review (Bosteels 2015).

Types of interventions

We included the following three randomised comparisons:

1. A routine screening hysteroscopy, with or without treatment of any detected uterine cavity abnormalities, versus no hysteroscopy, in subfertile women wishing to conceive spontaneously.

2. A routine screening hysteroscopy, with or without treatment of any detected uterine cavity abnormalities, versus no hysteroscopy, before intrauterine insemination (IUI).

3. A routine screening hysteroscopy, with or without treatment of any detected uterine cavity abnormalities, versus no hysteroscopy, before in vitro fertilisation (IVF).

Types of outcome measures

Electronic searches

We searched the following electronic databases, trial registers, and web sites from inception to 05 September 2018:

• Cochrane CGF Specialised Register, ProCite platform, searched 05 September 2018 (Appendix 1);

• Cochrane Central Register of Studies (CENTRAL CRSO), Web platform, searched 05 September 2018 (Appendix 2);

• MEDLINE, Ovid platform, searched from 1946 to 05 September 2018 (Appendix 3);

• Embase, Ovid platform, searched from 1980 to 05 September 2018(Appendix 4);

• PsycINFO, Ovid platform, searched from 1806 to 05 September 2018 (Appendix 5);

• CINAHL (Cumulative Index to Nursing and Allied Health Literature), EBSCO platform, searched from 1961 to 05 September 2018 (Appendix 6);

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

• World Health Organization International Trials Registry Platform (www.who.int/trialsearch/Default.aspx);

• DARE (Database of Abstracts of Reviews of Effects) on the Cochrane Library for reference lists from relevant non‐Cochrane reviews (onlinelibrary.wiley.com/o/cochrane/cochrane_cldare_articles_fs.html);

• Web of Knowledge (another source of trials and conference abstracts (wokinfo.com/));

• OpenGrey for unpublished literature from Europe (www.opengrey.eu/);

• LILACS (Latin American and Caribbean Health Science Information database) (regional.bvsalud.org/php/index.php?lang=en);

• PubMed and Google for recent trials not yet indexed in MEDLINE.

We combined the MEDLINE search with the Cochrane Highly Sensitive Search Strategy for identifying randomised trials, which appears in Section 6.4.11 of the Cochrane Handbook for Systematic Reviews of Interventions (Lefebre 2011).

We combined the Embase, PsycINFO, and CINAHL searches with trial filters developed by the Scottish Intercollegiate Guidelines Network (SIGN (www.sign.ac.uk/search‐filters.html)).

Searching other resources

Four review authors (JB, SS for non‐IVF comparisons; MSK, SKS for IVF comparison) handsearched reference lists of articles retrieved by the search, and contacted experts in the field to obtain additional data. We also handsearched relevant journals and conference abstracts that were not covered in the GFG register, in liaison with the Information Specialist.

Selection of studies

Two review authors (MSK, SKS) conducted an initial screen of titles and abstracts identified by the search, after which we retrieved the full texts of all potentially eligible studies. Four review authors (SS, JB, SKS, MSK) independently examined these full‐text articles for compliance with the inclusion criteria, and selected studies eligible for inclusion in the review. We corresponded with study investigators as required to clarify study eligibility. Disagreements as to study eligibility were resolved by discussion or by a fifth review author (BWJM). We documented the selection process with a PRISMA flow chart.

Data extraction and management

Four review authors (JB, MSK, SKS, SS) independently extracted data from eligible studies using a data extraction form designed and pilot‐tested by the review authors. Any disagreement was resolved by discussion or by a fifth review author (BWJM). We extracted data that included study characteristics and outcome data (Appendix 7). We corresponded with study investigators for further data on methods or results, or both, as required. We included studies irrespective of whether outcomes were reported in a 'usable' way.

Assessment of risk of bias in included studies

Four review authors (JB, SS for non‐IVF comparisons; MSK, SKS for IVF comparison) independently assessed the included studies for risk of bias using the Cochrane 'Risk of bias' assessment tool (Higgins 2011). We assessed the following items: selection (random sequence generation and allocation concealment); performance (blinding of participants and personnel); detection (blinding of outcome assessors); attrition (incomplete outcome data); reporting (selective reporting); and other bias. Disagreements were resolved by discussion or by a fifth review author. We described all judgements fully, and presented the conclusions in the 'Risk of bias' table, which were incorporated into the interpretation of review findings by means of sensitivity analyses (see Sensitivity analysis). Selective reporting is a type of reporting bias that affects the internal validity of an individual study. It refers to the selective reporting of some outcomes (e.g. positive outcomes) and the failure to report others (e.g. adverse events). We took care to search for within‐trial selective reporting, such as trials failing to report obvious outcomes, or reporting them in insufficient detail to allow inclusion. We compared the outcomes between the published protocol and the final published study. Where identified studies failed to report the primary outcome of live birth, but did report interim outcomes, such as clinical pregnancy, we undertook informal assessment as to whether the interim values (e.g. pregnancy rates) were similar to those reported in studies that also reported live birth.

Measures of treatment effect

For dichotomous data (e.g. live‐birth rates), we used the number of events in the control and intervention groups of each study to calculate risk ratios (RR). We used Peto odds ratio (OR) for outcomes with low event rates. We reversed the direction of effect of individual studies, if required, to ensure consistency across trials. We presented 95% confidence intervals (CI) for all outcomes. We compared the magnitude and direction of effect reported by studies with how they were presented in the current review, taking account of legitimate differences.

Unit of analysis issues

The primary analysis was per woman randomised; we included per pregnancy data for some outcomes (for the outcome miscarriage). We counted multiple live births (e.g. twins or triplets) as one live‐birth event.

Dealing with missing data

We analysed the data on an intention‐to‐treat basis to the greatest degree possible, and attempted to obtain missing data from the original authors. We assumed live births or clinical pregnancies would not be present in women without a reported outcome. For other outcomes, we analysed only the available data.

For Imputated data, we had planned to conduct sensitivity analysis (see Sensitivity analysis).

Assessment of heterogeneity

We considered whether the clinical and methodological characteristics of the included studies were sufficiently similar for meta‐analysis to provide a clinically meaningful summary. We assessed statistical heterogeneity using the I² statistic. An I² measurement greater than 50% indicated substantial heterogeneity (Higgins 2011).

Assessment of reporting biases

In view of the difficulty of detecting and correcting for publication bias and other reporting biases, the review authors aimed to minimise the potential impact of these biases by ensuring a comprehensive search for eligible studies, and by being alert for duplication of data. Since there were fewer than 10 studies in each population, we did not use a funnel plot.

Data synthesis

Four review authors (JB, SS for non‐IVF comparisons; MSK, SKS for IVF comparison) entered the data and performed the statistical analysis using Review Manager 5 (RevMan 2014). We combined the data using a fixed‐effect model for the following comparisons.

• Hysteroscopy versus no hysteroscopy for subfertile women wishing to conceive spontaneously

• Hysteroscopy versus no hysteroscopy for subfertile women undergoing IUI

• Hysteroscopy versus no hysteroscopy for women undergoing IVF

We displayed an increase in the odds of a particular outcome, which may be beneficial (e.g. live birth) or detrimental (e.g. adverse effects of the hysteroscopy) graphically in the meta‐analyses to the right of the centre‐line, and a decrease in the odds of an outcome to the left of the centre‐line.

Subgroup analysis and investigation of heterogeneity

We conducted the following subgroup analysis:

For women undergoing IVF:

• First IVF versus two or more IVF failures.

Sensitivity analysis

We conducted sensitivity analyses for the primary outcome live birth and an important secondary outcome (clinical pregnancy rate) to determine whether the conclusions were robust to arbitrary decisions made regarding eligibility and analysis. These analyses included consideration of whether the review conclusions would have differed if:

• eligibility had been restricted to studies without high or unclear risk of bias in any domain;

• a random‐effects model had been adopted;

• alternative imputation strategies had been implemented;

• the summary effect measure had been odds ratio rather than risk ratio;

• the primary outcome had been limited to live birth.