Tap and inject of intravitreal antibiotics versus pars plana vitrectomy for post-cataract surgery endophthalmitis: a meta-analysis

Author: Mihalache A, Patil NS, Popovic MM, Sedrak P, Kertes PJ, Muni RH

Geographical coverage: India, Iran, Saudi Arabia, Taiwan, Greece and the Netherlands

Sector: Cataract surgery

Sub-sector: Efficacy and safety

Equity focus: Not reported

Study population: Patients with cataract

Review type: Effectiveness review

Quantitative synthesis method: Meta-analysis

Qualitative synthesis method: Not applicable

Background:

Post-operative bacterial endophthalmitis is an uncommon but sight-threatening complication of cataract surgery, with an incidence of 0.02 %–0.26 %. As surgical volumes rise worldwide, optimising management is increasingly important. Initial treatment options include intravitreal tap–and-inject (T/I) of broad-spectrum antibiotics and primary pars plana vitrectomy (PPV). Guidance from the 1995 Endophthalmitis Vitrectomy Study (EVS) recommended PPV only for eyes presenting with light-perception vision, but advances in microsurgical techniques and antibacterial therapy necessitate a re-evaluation of relative benefits and risks.

Objective:

To compare the efficacy and safety of initial T/I with those of PPV for post-cataract-surgery bacterial endophthalmitis.

Main findings:

Seven comparative studies (three prospective, four retrospective; 188 eyes) met the inclusion criteria. Risk of bias for all non-randomised studies was rated ‘high’; certainty of evidence for key outcomes was very low.

• Visual acuity: Eyes managed with initial T/I achieved better final best-corrected visual acuity (BCVA) than those treated with primary PPV (weighted mean difference = –0.61 logMAR; 95 % CI –1.19 to –0.03; p = 0.04).
• Safety: No significant differences were detected between T/I and PPV for enucleation/evisceration (risk ratio [RR] = 0.73; 95 % CI 0.09–6.25) or retinal detachment (RR = 0.29; 95 % CI 0.01–5.94).
• Heterogeneity & bias: Substantial clinical heterogeneity, small sample sizes and confounding limit confidence in pooled estimates.

Overall, very low-certainty evidence suggests that initial T/I may yield superior visual outcomes without increasing serious adverse events compared with PPV, but robust randomised trials are required to confirm these findings.

Methodology:

MEDLINE, Embase and the Cochrane Library were searched for English-language comparative studies (January 1990 – January 2021) reporting BCVA after T/I or PPV for post-cataract-surgery infectious endophthalmitis. Two reviewers independently screened records, extracted data and assessed quality using ROBINS-I; disagreements were resolved by consensus or a third reviewer. Random-effects meta-analyses generated pooled estimates; statistical heterogeneity was quantified with I². Sensitivity analyses evaluated the influence of individual studies. Evidence certainty was appraised with GRADE.

Applicability / external validity:

Generalisability is restricted by small study cohorts, high risk of confounding, geographical concentration in Asia and Europe, and evolving surgical/antibiotic practice since the study period. Findings should therefore be interpreted with caution until larger, multicentre randomised trials provide higher-quality evidence.

Geographic focus:

Included studies were conducted in India, Iran, Saudi Arabia, Taiwan, Greece and the Netherlands.

Summary of quality assessment:

There is medium confidence in the review’s conclusions: searches were comprehensive, eligibility criteria explicit, and dual-reviewer methods employed. However, English-language restriction, lack of an excluded-studies list and uniformly high risk of bias in included studies reduce overall certainty.