Types of studies

We included only randomized controlled trials (RCTs), irrespective of their publication status or language.

Types of participants

We included trials that enrolled participants with either unilateral or bilateral GRT. We define GRT as a full‐thickness retinal break extending circumferentially for 90° or greater in the presence of a posterior vitreous detachment, or as defined by eligible trials. We included trials with more than three months of follow‐up, regardless of age, gender, ethnicity, lens status (e.g. phakic or pseudophakic eyes) of the affected eye(s), or etiology of GRT among participants enrolled in the trials.

Types of interventions

We included trials that compared PPV combined with scleral buckle versus PPV alone for GRT.

Types of outcome measures

Electronic searches

The Cochrane Eyes and Vision Information Specialist searched the following electronic databases for RCTs and controlled clinical trials. There were no restrictions on language or year of publication. Electronic databases were last searched on 16 August 2018.

• Cochrane Central Register of Controlled Trials (CENTRAL; 2018, Issue 8) (which contains the Cochrane Eyes and Vision Trials Register), in the Cochrane Library (searched 16 August 2018) (Appendix 1).

• MEDLINE Ovid (1946 to 16 August 2018) (Appendix 2).

• Embase.com (1947 to 16 August 2018) (Appendix 3).

• PubMed (1948 to 16 August 2018) (Appendix 4).

• Latin American and Caribbean Health Science Information Database (LILACS) (1982 to 16 August 2018) (Appendix 5).

• US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (www.clinicaltrials.gov; searched 16 August 2018) (Appendix 6).

• World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) (www.who.int/ictrp; searched 16 August 2018) (Appendix 7).

Searching other resources

We searched the reference lists of reports from trials we identified, to look for additional trials. We did not conduct manual searches of conference proceedings or abstracts specifically for this review. We used the Science Citation Index to find studies that cited the identified trials (last searched 21 July 2019).

Selection of studies

Two review authors independently assessed the titles and abstracts of all records identified by electronic and manual searches. Each review author labeled the study referenced in each citation as "definitely relevant," "unsure," or "definitely not relevant." We excluded from the review trials classified as "definitely not relevant." Two review authors independently reassessed the full text of trials labeled as "unsure" and "definitely relevant" according to the inclusion criteria for this review and classified them as "definitely include" or "definitely exclude." We assessed trials labeled as "definitely include" for methodological quality. We resolved any differences in classification between the two review authors through discussion at both stages of the screening process. We documented excluded trials after review of the full‐text reports and provided reasons for exclusion in the Characteristics of excluded studies table.

Data extraction and management

Two review authors extracted data independently using Covidence. Two review authors reached consensus before entering data into Review Manager 5 (RevMan 5) (Review Manager 2014). We collected the following information from the included trials: study methods, participants, interventions, and outcomes. We contacted the authors of two conference abstracts that were not published, to ask for more information about the studies; we did not receive any data after two weeks.

Assessment of risk of bias in included studies

We assessed each included trial for risks of bias according to the guidelines provided in Chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2017). We considered the following criteria when assessing bias.

• Random sequence generation: we planned to assess the method used to generate the allocation sequence generation and the allocation concealment method used before randomization.

• Masking (blinding) of participants and outcome assessors: we planned to assess the methods used to mask participants and outcome assessors.

• Incomplete outcome data: we planned to assess exclusions after randomization, rates of follow‐up, reasons for losses to follow‐up, and deviations from intention‐to‐treat analysis of outcomes.

• Selective outcome reporting: we planned to assess selective outcome reporting by comparing protocols and other design documents with trial reports and by noting outcomes that were measured but not reported.

• Other bias: we planned to assess other sources of bias, for example, industry funding.

We graded each trial on the basis of the above bias criteria as:

• high risk of bias: when plausible bias seriously weakens the study results;

• low risk of bias: when plausible bias is unlikely to alter the study results; or

• unclear risk of bias: when plausible bias raises some doubt about the study results.

For all trials graded as "unclear risk of bias" because of doubt about the study on methods, results, or missing data, we contacted the investigators of the RCTs to request additional information. We assessed risks of bias of using available information if we did not hear back from trial investigators in two weeks, and we noted lack of response from trial investigators. We resolved any disagreements between review authors regarding bias assessment through discussion until we reached consensus.

Measures of treatment effect

We estimated the risk ratio (RR) and its 95% confidence interval (CI) after surgery (PPV combined with scleral buckle vs PPV alone) for the following dichotomous outcomes with information obtained from the included studies.

• Primary surgical success.

• Second surgery for retinal reattachment.

• Development of adverse events such as retinal detachment recurrence, elevation of intraocular pressure above 21 mmHg, choroidal detachment, cystoid macular edema, macular pucker, proliferative vitreoretinopathy, progression of cataract in initially phakic eyes, and any other adverse events reported by included trials at any time from day one up to the last reported follow‐up visit after surgery.

We planned to estimate means and standard deviations of absolute and relative change after surgery (PPV combined with scleral buckle vs PPV alone) for the following outcomes.

• Absolute and relative change in visual acuity in logMAR from baseline to follow‐up visits in both surgery groups.

• Absolute and relative change in retinal detachment in GRT from baseline to follow‐up visits in both surgery groups.

To avoid the phenomenon of regression to the mean for the absolute change, we planned to perform an analysis of covariance where the change was defined relative to the value of "y" at baseline.

Unit of analysis issues

For this review, the unit of analysis was the eye; all statistical estimations included within‐individual correlation among analyses. We planned to use covariance and linear regression for repeated measures.

For trials that enrolled both eyes of participants with bilateral GRT and used a paired‐eye design in which one eye was randomized to receive PPV alone and the fellow eye received the combined procedure, the unit of analysis was the eye. If enough data were provided for paired‐eye design trials such as covariance or paired tabulation, we planned to use a paired‐sample t‐test or a McNemar test to assess differences between paired‐eye outcomes.

Dealing with missing data

If trials were not analyzed via an intention‐to‐treat approach, or when data were unclear or missing, we planned to contact trial authors for clarification and further information. We allowed two weeks for trial authors to respond, and if we did not receive a response, we used available information.

Assessment of heterogeneity

We planned to evaluate the heterogeneity of trials by using Cochrane Q and I² tests. The Q value has a Chi² distribution with the null hypothesis that the likelihood ratio is the same for all studies. The I² test evaluates variability of effect. We planned to consider values higher than 50% as evidence of substantial heterogeneity (Deeks 2017). If substantial heterogeneity or inconsistency was present, we did not report the pooled analysis; we planned to report instead a narrative summary of the results.

We expected methodological and clinical sources of heterogeneity from differences in participant characteristics, types or timing of outcome measurements, and intervention characteristics such as surgical methods used, including number of surgeons, PPV gauge number, scleral buckle type (silicon band/silicon sponge), preoperative lens status (phakic, pseudophakic, aphakic), and vitreous substitute endotamponade (gas/silicon oil).

Assessment of reporting biases

When possible, we planned to obtain published protocols or methods papers to compare intended outcomes versus reported outcomes. When we identified 10 or more RCTs, we planned to use funnel plots to judge asymmetry that may indicate publication bias in one or more reported studies in a meta‐analysis.

Data synthesis

We planned to perform data analysis according to the guidelines set forth in Chapter 9 of the Cochrane Handbook for Systematic Reviews of Interventios (Deeks 2017). We planned to compare PPV combined with scleral buckle versus PPV alone for GRT. When the number of trials was less than three, we planned to use a fixed‐effect model; otherwise we planned to to use a random‐effects model. If we found evidence of substantial heterogeneity, or if we noted large diversity in participant characteristics and trial methods, we planned to present only a narrative summary of the results. We planned to perform a meta‐analysis for all outcomes under Types of outcome measures.

Subgroup analysis and investigation of heterogeneity

We planned to investigate potential explanations for clinical or statistical heterogeneity by comparing outcomes within subgroups to compare the relative effects of PPV combined with scleral buckle and PPV alone by type of tamponade used in surgery (silicon oil or gas), participant age, retinal tear etiology, duration of symptoms, grades or quadrants of retinal tear extension (90° to 180° or one or two quadrants, 180° to 270° or two or three quadrants, and 270° to 360° or three or four quadrants), lens status (phakic, pseudophakic, or aphakic), and presence of proliferative vitreoretinopathy (grade C) according to the Retinal Society classification (Machemer 1991), whenever sufficient data were reported.

Sensitivity analysis

We planned to examine the impact of excluding trials with high risk of bias in all of the following domains ‐ incomplete data, unpublished data, and industry funding ‐ to assess the robustness of estimates with respect to these factors.