Outcomes of pars plana vitrectomy following ocular trauma at varying surgical time points

Abstract

Introduction
The optimal timing of pars plana vitrectomy (PPV) following ocular trauma is an ongoing debate. Early vitrectomy post-trauma enables the rapid assessment of occult retinal pathology. On the other hand, late vitrectomy is less challenging as there is a lower risk of bleeding and posterior vitreous detachment induction is easier. The purpose of this work is to report the functional and anatomical outcomes following ocular traumatic injuries in a US-based cohort, emphasizing the time of intervention.
Methods
This was a retrospective case series of 110 patients with traumatic ocular injuries who underwent PPV from 2008 to 2020. Patients were grouped into four timing categories: same day (0 days), early (1-7 days), delayed (8-14 days), and late (>14 days). Multivariable regression models controlling for confounding were implemented to assess the impact of vitrectomy timing on anatomical and functional outcomes. Visual acuity (VA) at baseline and after surgery, proliferative vitreoretinopathy (PVR), and enucleation for each vitrectomy timing category.
Results
Patient demographics and severity of ocular trauma were comparable across timing categories. Final VA in logMAR was found to have a stepwise worsening as the time of ocular trauma to vitrectomy was increased (P < 0.05). For every one-step increase in the vitrectomy timing category, there was an adjusted 0.24 (CI, 0.04-0.44) increase in final VA. No patient in the same day vitrectomy group had an enucleation or PVR, while patients who had late vitrectomies had the largest number of both enucleations and PVR (44.4% and 52.0%, resp.). In adjusted analysis, there was 3.11 increased odds (CI, 1.03-9.42) of developing PVR for a one-step increase in vitrectomy timing (P < 0.05).
Conclusions
Vitrectomy on the same day of injury has the best final VA, and the lowest incidence rates of PVR and enucleation in comparison to other timing categories, regardless of etiology.