OASIS 2016.

Methods	Study design: RCT, single treated eye. Number randomised: 220 total; 146 ocriplasmin; 74 sham. Exclusions after randomisation: 50 participants subsequently deemed ineligible by central reading centre. Number analysed: at 28 days: 168 total; 111 ocriplasmin; 59 sham. Unit of analysis: eyes. Losses to follow‐up: 2 participants total; 1 ocriplasmin group (1 lost to follow‐up); 1 sham group (1 lost to follow‐up). How was missing data handled? other than VMA release at 28 days, no data published regarding cohort who met central reading centre eligibility criteria. Power calculation: 210 participants for at least 90% power at 2‐sided alpha of 0.05 to assume a primary endpoint of 37% in ocriplasmin group and a 14% rate in placebo group.
Participants	Country: USA. Mean age: 69.1 years overall; 69.4 years for ocriplasmin group; 68.5 years for sham group. Gender: 147/218 (67.4%) women, 71/218 (32.6%) men total; 102/145 (70.3%) women, 43/145 (29.7%) men in ocriplasmin group; 45/73 (61.6%) women, 28/73 (38.4%) men in sham group. Inclusion criteria: aged > 18 years; presence of VMA; BCVA ≤ 20/32 in study eye; BCVA ≥ 20/800 in non‐study eye. Exclusion criteria: history or current evidence of proliferative retinopathy, exudative AMD or retinal vein occlusion in the study eye; people with any vitreous haemorrhage or any other vitreous opacification which precludes either visualisation of the posterior pole by visual inspection OR adequate assessment of the macula by OCT; MH > 400 µm in diameter in the study eye; presence of epiretinal membrane; aphakia in study eye; high myopia (> ‐8 dioptres in study eye); history of rhegmatogenous retinal detachment in either eye; prior vitrectomy in study eye; previous participation in this trial or prior administration of ocriplasmin in study eye. Equivalence of baseline characteristics: yes.
Interventions	Intervention: single intravitreal injection of ocriplasmin 125 µg in 0.10 mL volume. Comparator: sham (the same syringe hub was pressed against the conjunctiva to simulate an injection). Length of follow‐up: planned 24 months, actual 24 months. Data of central reading centre approved study participants only reported at 28 days.
Outcomes	Primary outcome, as defined in study reports: "proportion of subjects with pharmacological vitreomacular adhesion (VMA)/vitreomacular traction (VMT) resolution at day 28. Pharmacological VMA resolution without anatomical defect, based on spectral domain optical coherence tomography and determined by the masked central reading center (CRC), with post‐resolution vitrectomy considered as a failure." Secondary outcomes, as defined in study reports: "proportion of subjects with a ≥2 line improvement in best‐corrected visual acuity (BCVA) from baseline at month 24, irrespective of vitrectomy." Adverse events reported: yes. Intervals at which outcomes assessed: 7 and 28 days; 3, 6, 9, 12, 15, 18, 21 and 24 months.
Notes	Funding sources: study sponsored by ThromboGenics NV. Study period: 3 years; 2011‐2014. Reported subgroup analyses: yes. Additional information: large proportion of participants were deemed eligible, recruited and treated by investigators. Retrospective central reading centre review found 50 participants ineligible for following reasons (MH > 400 μm, presence of epiretinal membrane or no sVMA at baseline). Our analysis only included data reported for correctly eligible cohort of participants. NCT01429441.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Method of random sequence generation for the MIVI‐IIT RCT described Quote: "Randomization was stratified on the basis of the presence or absence of FTMH at baseline and was centralized through an interactive voice response system." p. 2233.
Allocation concealment (selection bias)	Low risk	Method of allocation concealment for MIVI‐IIT RCT described. Quote: "Randomization was stratified on the basis of the presence or absence of FTMH at baseline and was centralized through an interactive voice response system." p. 2233.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Performance bias explained. Quote: "The trial was conducted in a double‐masked manner. To maintain the masking of the investigator, an unmasked injecting physician was assigned to perform the injection and access the interactive voice response system to receive the assigned treatment. The unmasked personnel did not perform or participate in any other trial‐related procedures or assessments." p. 2233.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Detection bias appropriately explained. Quote: "The trial was conducted in a double‐masked manner. To maintain the masking of the investigator, an unmasked injecting physician was assigned to perform the injection and access the interactive voice response system to receive the assigned treatment. The unmasked personnel did not perform or participate in any other trial‐related procedures or assessments." p. 2233.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Large proportion of participants deemed eligible, recruited and treated by investigators. Retrospective central reading centre review found 50 participants ineligible for following reasons (MH > 400 μm, presence of epiretinal membrane or no sVMA). Outcome data for correct eligible cohort of participants only given for primary outcome. No secondary outcome data described.
Selective reporting (reporting bias)	Low risk	All outcomes defined in trial registry were reported.
Other bias	Low risk	No other source of bias.