Albrecht 2011.

Methods	Study design: randomized controlled trial Number randomized: 62 participants in total, 32 in postcataract endophthalmitis group (17 steroid, 15 placebo), 13 in bleb‐related endophthalmitis group (4 steroid, 9 placebo), 17 in other‐endophthalmitis group (9 steroid, 8 placebo) Exclusions after randomization: admission visual acuity was not recorded for 1 participant in other‐endophthalmitis group who received steroids Losses to follow‐up: 1 participant in postcataract group placebo group, 2 participants in bleb‐related placebo group, 1 participant in other placebo group Number analyzed: 57 participants in total, 31 in postcataract group (17 steroid, 14 placebo), 11 in bleb‐related group (4 steroid, 7 placebo), 15 in other group (8 steroid, 7 placebo) Unit of analysis: participant (1 eye per participant) Notes: other‐endophthalmitis group included 8 trauma‐related (4 steroid, 4 placebo), 3 endogenous (1 steroid, 2 placebo), and 5 post‐pars plana vitrectomy endophthalmitis (4 steroid, 2 placebo) participants
Participants	Country: South Africa Setting: Groote Schuur Hospital (Cape Town) Study period: January 2001 to December 2005 Mean age (years): 59 (steroid), 61 (placebo) Gender: 11 men and 19 women (steroid); 18 men and 14 women (placebo) Inclusion criteria: all people with presumed bacterial endophthalmitis Exclusion criteria: suspected fungal/parasitic/viral/non‐bacterial endophthalmitis people who underwent vitrectomy for endophthalmitis Equivalence of baseline characteristics: yes
Interventions	Intervention (n = 30): intravitreal dexamethasone 0.4 mg/1 mL with intravitreal vancomycin 1 mg/0.1 mL and ceftazidime 2.225 mg/0.1 mL (replaced with amikacin 0.4 mg/1 mL for participants allergic to penicillin) Comparator (n = 32): intravitreal placebo 0.1 mL balanced salt solution with intravitreal vancomycin 1 mg/0.1 mL and ceftazidime 2.225 mg/0.1 mL Additional interventions (all participants): "Vitreous and aqueous samples were sent for microbiological analysis. A subconjunctival injection of vancomycin (25 mg/0.5 ml), ceftazidime (50 mg/0.5 ml) and betamethasone (1.5 mg/0.5 ml) was also administered at the end of the procedure. Post injection, patients received topical ofloxacin and topical dexamethasone. Patients were re‐injected after 48‐72 h if needed." Length of follow‐up: 2 to 4 months
Outcomes	Outcomes: visual acuity using standard Snellen chart, grouped into the following categories (group 1: 6/6 to 6/18, group 2: 6/24 to 6/60, group 3: < 6/60) number of lines improvement on the Snellen chart any adverse events any medication side effects Other findings reported: % positive culture rate overall, most common organism cultured, mean delay in presentation of endophthalmitis Adverse events reported: rhegmatogenous retinal detachment Intervals at which outcomes were assessed: baseline, 3 months
Notes	References to other relevant studies: none Trial registration: none reported Funding source: none reported Declarations of interest: none reported
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"The pharmacy randomised the patients within the three groups using standard computer generated randomisation tables."
Allocation concealment (selection bias)	Low risk	Allocation was performed by the pharmacy.
Masking of participants and personnel (performance bias)	Low risk	"A double‐blinding label (dexamethasone/placebo) masked the dexamethasone/placebo injection to both surgeon and patient."
Masking of outcome assessment (detection bias)	Low risk	Visual acuities were measured by nursing staff who had no knowledge of the trial, so they were unaware of the participants' intervention assignments. Note: this information was provided by the study author via email.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Out of the total 62 participants, 4 participants in the placebo group were lost to follow‐up and the admission visual acuity was not recorded for a participant in the steroid group. The reasons for losses to follow‐up were not reported (and unknown to the study author when asked by email). These participants were excluded from the final analysis (after randomization).
Selective reporting (reporting bias)	Unclear risk	We did not have access to the original study protocol to compare planned versus reported outcomes.
Other bias	Low risk	None identified.