Abstract
OBJECTIVES: To assess the influence of drainage retinotomy (DrR) on anatomical and visual outcomes of pars plana vitrectomy (PPV) for primary uncomplicated rhegmatogenous retinal detachment (RD), compared with drainage through pre-existing retinal break (PRB). METHODS AND ANALYSIS: Retrospective study on patients treated with PPV for RD. Prospectively collected data were extracted from the Britain & Eire Association of Vitreoretinal Surgeons and European Society of Retina Specialists (EURETINA) RD database, including baseline features, surgical details, and anatomical and functional outcomes. Inclusion criteria were as follows: uncomplicated PPV, gas tamponade, drainage through DrR or PRB, surgeons with >100 cases recorded. Exclusion criteria were as follows: age <16, <2-month follow-up, ocular comorbidity, proliferative vitreoretinopathy ≥grade C, giant retinal tear, tamponade other than gas. Full propensity score matching resulted in matched groups to mitigate confounding bias. Subsequent multivariable linear regression was performed for postoperative best-corrected visual acuity (BCVA) as dependent variable, and Firth penalised logistic regression with DrR, single-surgery anatomical success (SSAS), epiretinal membrane (ERM) and macular fold as dependent dichotomised variables on matched data. RESULTS: Of 12 504 eyes extracted, 4175 were included. Of these, 3432 (82.2%) had PRB drainage (non-DrR group) and 743 (17.8%) a DrR (DrR group). Final median (IQR) BCVA was 0.18 (0.14-0.48) in the non-DrR group and 0.20 (0.18-0.48) in the DrR group (p=0.072). SSAS rate was 93.4% and 91% (OR 0.71 (95% CI 0.54 to 0.95)) and postoperative ERM rate 1.6% and 4.2% (OR 2.63 (95% CI 1.68 to 4.10)) in the non-DrR and DrR groups, respectively. On multivariable regression, DrR was associated with postoperative ERM (p=0.011), but not with final BCVA, SSAS and macular folds (p=0.633, 0.149 and 0.085, respectively). CONCLUSION: Our study confirmed the association between DrR and increased risk of developing ERM; however, DrR does not appear to impact significantly on other outcomes.