Timing of vitrectomy for treatment of endophthalmitis after intravitreal anti-VEGF injection: a systematic literature review of case reports and series

Abstract

Objective:

To perform a systematic literature review analyzing visual outcomes of immediate, early, and delayed vitrectomy in the treatment of acute endophthalmitis after intravitreal anti-vascular endothelial growth factor (anti-VEGF) injections.

Methods:

We conducted a literature search using the Ovid Medline, Embase.com, and Web of Science databases, and relevant articles were selected from original English papers published from 2005 to 2021. Inclusion criteria were studies reporting cases of acute post-anti-VEGF endophthalmitis, defined as occurring within 6 weeks of injection treatment. Exclusion criteria were pediatric cases and cases explicitly reported to be caused by injections of contaminated drugs. Risk of bias was assessed using the Joanna Briggs Institute Critical Appraisal tool for case reports and case series. The study dataset for descriptive and statistical analysis comprised patient-level data extracted from included studies. The timing of vitrectomy compared were defined as (1) immediate vitrectomy as occurring within 24 h of endophthalmitis diagnosis; (2) early vitrectomy as occurring between 24 and 48 h of endophthalmitis diagnosis; (3) late vitrectomy as occurring after 48 h of endophthalmitis diagnosis. Primary outcome was final visual acuity following treatment with vitrectomy.

Results:

Twenty-five articles were published that met our inclusion and exclusion criteria for a total of 86 cases. Thirty-seven were immediate vitrectomy, 25 were early, and 24 were late vitrectomy treatment groups, respectively. We observed differences in final visual outcomes and in improvement from diagnosis to final visual acuity, with patients receiving immediate and late vitrectomy to have better final visual outcomes than those patients receiving early vitrectomy (p < 0.005).

Conclusion:

Our results show that there may be an association between time to vitrectomy and visual outcomes. Immediate and late vitrectomy treatment groups had better visual outcomes than the early group. Our results were limited by the reliance on case reports and series and the paucity of data available specifying the timing of vitrectomy. Additional research is necessary to elucidate the effects of treatment timing in patients with endophthalmitis following anti-VEGF injection.

Plain language summary

Timing of vitrectomy for treatment of endophthalmitis after intravitreal anti-VEGF injection: a systematic literature review of case reports and series

In this study, we aimed to review existing literature to assess the associations between timing of surgery to treat endophthalmitis following intravitreal anti-VEGF injection and visual outcomes. The timing of vitrectomy compared were defined as (1) immediate vitrectomy as occurring within 24 hours of endophthalmitis diagnosis; (2) early vitrectomy as occurring between 24-48 hours of diagnosis; (3) late vitrectomy as occurring after 48 hours of diagnosis. We found 25 articles with 86 total cases. We observed differences in final visual outcomes and in improvement from diagnosis to final vision, with patients receiving immediate and late vitrectomy to have better final vision than those patients receiving early vitrectomy.

Introduction

Intravitreal injections of anti-vascular endothelial growth factor (anti-VEGF) medications come with a risk of endophthalmitis ranging from 0.049% to 0.056%.^1,2 Treatment options for endophthalmitis include a needle vitreous biopsy accompanied by intravitreal injection of antibiotics (tap and inject; TAI) or pars plana vitrectomy (PPV) accompanied by intravitreal antibiotics. To date, the Endophthalmitis Vitrectomy Study (EVS) published in 1995 is the only randomized controlled trial to compare TAI and PPV for the treatment of post-surgical endophthalmitis. The EVS reported equivalency between TAI and PPV as initial treatments except for eyes presenting with a vision (VA) of light perception (LP), in which PPV was superior.

Since the EVS, the epidemiology of endophthalmitis has been changing.^3,4 With the advent of anti-VEGF as a mainstay treatment for vitreoretinal pathologies, the incidence of endophthalmitis due to intravitreal injections has been increasing. Furthermore, the disease course may vary due to the causative intraocular flora introduced with injection as compared with surgery. ⁵

In the context of ongoing improvements in vitrectomy techniques, tools, and outcomes, it is important to continually evaluate the appropriate management of this disease. ⁶ There have been no published controlled trials to compare the treatment groups for postinjection endophthalmitis, and clinical practice is largely guided by the results from the EVS.^7,8 Following the EVS, multiple retrospective studies and reports evaluating best practices for the management of postinjection endophthalmitis have been inconclusive regarding standardization of care.^6,9–15 To date, there is one systematic review comparing PPV and TAI as the initial treatments of endophthalmitis following anti-VEGF injections, which found no significant difference in visual outcomes between the treatment groups in published cases. ¹³

No study to date has evaluated vitrectomy timing in the treatment of endophthalmitis following anti-VEGF injections. The purpose of this systematic review is to compare the visual outcomes of cases of endophthalmitis occurring after anti-VEGF injection treated with immediate, early, and late vitrectomy. The timing of vitrectomy compared were defined as (1) immediate vitrectomy as occurring within 24 h of endophthalmitis diagnosis; (2) early vitrectomy as occurring between 24 and 48 h of endophthalmitis diagnosis; (3) late vitrectomy as occurring after 48 h of endophthalmitis diagnosis.

Methodology

Study identification and eligibility criteria

We conducted a structured systematic review of the literature to identify all published reports pertaining to the outcome of endophthalmitis after intravitreal injections in which the treatment included vitrectomy. The PRISMA 2020 checklist for this study may be found in Supplemental Material 1. Studies were identified through a search using Ovid Medline (1946–2021), Embase.com (1947–2021), and Web of Science (Core Collection 1900–2021) database to find original English articles with the keywords “endophthalmitis” AND “intravitreal injections” OR “intravitreal” OR “drug or administration or injection” OR “antibiotic” OR “bevacizumab” AND “vitrectomy” OR “vitreous resection.” Keywords were selected based on expert opinion (M.G.K.) and evaluation of selected review articles on post-intravitreal injection endophthalmitis and vitrectomy. The search was completed on November 23, 2021.

Published articles including case reports, case series, or studies were considered eligible for our review if they satisfied our inclusion criteria: (1) the article was written in English; (2) the study examined human cases; (3) studies after the year 2005 (unless they are clinical trials for anti-VEGF drugs that included patients who developed endophthalmitis); (4) the article contained case(s) of acute endophthalmitis after intravitreal anti-VEGF injections; (5) visual acuity was documented for the patient(s) at the time of endophthalmitis diagnosis; (6) follow-up visual acuity was taken; (7) papers in which the patient(s) had sterile endophthalmitis in which the patient(s) was treated; (8) cases in which the patient(s) was treated for their endophthalmitis with a vitrectomy and TAI of intravitreal antibiotics or vitrectomy alone. We defined acute endophthalmitis as endophthalmitis occurring within 6 weeks of an intravitreal injection. This definition was based on the EVS study which included cases with “clinical signs and symptoms of bacterial endophthalmitis within six weeks after cataract surgery or secondary intraocular lens implantation.” The decision to include clinically diagnosed cases, including both culture-positive and negative, if tested, allowed for increased power for the analysis of this relatively rare complication.

Articles were excluded from this review if they met any of the following exclusion criteria: (1) pediatric cases; (2) paper that included case(s) of endophthalmitis after injections of tainted drugs, contaminated drugs related to compounding pharmacies endophthalmitis outbreaks, antivirals, or steroids.

Study selection and data collection

Following the computerized search, we reviewed the titles and abstracts of all retrieved citations. Articles were rejected after review of abstracts only if it could be determined that they did not meet the inclusion criteria or if any of the exclusion criteria applied. We obtained and reviewed full copies of the articles in detail to determine whether the study satisfied the selection criteria.

Covidence (Veritas Health Innovation, Melbourne, VIC, Australia, 2022) was used for article screening. Two study authors (D.J.H. and S.G.) independently reviewed titles, abstracts, and full texts to determine article inclusion in accordance with predetermined criteria. The third reviewer (M.G.K.) resolved conflicts during the screening process.

We sent emails requesting patient-level data to the corresponding authors of studies for whom our study variables were not explicit in the published articles. Follow-up emails were sent at two time points if no response was received. For the responding corresponding authors, we sent a data collection form including the study variables listed below.

A full detailed selection process of the studies can be found in Figure 1.

Figure 1.

Study selection.

*The corresponding authors for 76 studies were emailed, of which 6 provided additional patient-level data for inclusion in the study. The remaining 70 studies were excluded due to incomplete extractable patient-level data.

Articles that satisfied the eligibility criteria underwent independent structured review by two authors (D.J.H. and S.G.) to extract study data. A third reviewer (M.G.K.) resolved conflicts during the data collection process. Study data collected included: (1) Covidence ID #; (2) study ID; (3) year of publication; (4) date of publication (years); (5) first author name; (6) contact information for the corresponding author; (7) country in which the study conducted; (8) type of study; (9) number of participants included; (10) results by patient. For each patient, the following was recorded: (1) patient number; (2) age; (3) gender; (4) race, (5) primary diagnosis for which intravitreal injection has been performed; (6) intravitreal injection drug injected before endophthalmitis diagnosis; (7) treatment for endophthalmitis; (8) visual acuity before diagnosis; (9) when vision was taken before diagnosis; (10) time between injection and presentation with symptoms; (11) visual acuity at the time of diagnosis; (12) when vitrectomy performed after diagnosis; (13) final visual acuity after diagnosis; (14) when final visual acuity was measured after diagnosis; and (15) culture data, if available. Patient data was only recorded for cases that met our selection criteria.

Quality assessment

All included studies were case reports or case series. Data was collected on the level of the individual patient rather than cohort. As such, quality assessment was performed using the Joanna Briggs Institute (JBI) critical appraisal tool for case reports and case series.^16,17 Two study authors (D.J.H. and S.G.) independently graded included studies in alignment with the JBI tool, and conflicts were resolved with the third reviewer (M.G.K.).

Data synthesis and statistical analysis

Because of the inclusion of case reports and series in the systematic review, we further guided our data synthesis by the Synthesis Without Meta-analysis (SWiM) guideline, which was created to supplement PRISMA to offer a more detailed structure for standard reporting of methods and results. ¹⁸ The SWiM checklist may be found in Supplemental Material 1. As no single study included offered enough data for statistical analysis of adequate power, a common dataset was created, synthesized, and analyzed.

For synthesis and analysis, individual cases were allocated to three treatment groups as defined above: (1) immediate vitrectomy, (2) early vitrectomy, and (3) late vitrectomy. The Snellen visual acuity was converted to logMAR equivalent for statistical analysis. ¹⁹ As established by prior studies, vision levels of count fingers (CF), hand motion (HM), LP, and no LP were assigned visual acuity values of 2.1, 2.4, 2.7, and 3.0, respectively. ²⁰

For the primary outcome, VA at baseline, diagnosis, and reported final were compared and presented. Changes in VA were analyzed per patient across the three time points by evaluating mean difference and assessing the magnitude of change to account for heterogeneity in baseline and presenting VA.

Descriptive statistics were used to summarize all data collected. Statistical methods for analyses and assessment of heterogeneity were performed including the t test analysis of variance for continuous outcomes, and the Chi-squared test for categorical outcomes, with statistical significance defined as a p-value of less than 0.05. All descriptive and statistical analyses and data visualization were performed in Microsoft^® Excel^® (2024).

Results

Study characteristics

Our search strategy identified 1,246 studies for title and abstract screening. We reviewed the full articles for 287 studies. A total of 19 studies met the inclusion criteria and had our study variables available in the manuscript. We emailed the corresponding authors for an additional 76 studies with incomplete extractable patient-level data, of which 6 studies provided patient-level data. A total of 25 studies were included in the study for review and analysis. Further information regarding the selection of studies can be found in Figure 1.

Our systematic search returned only case reports, case series, and retrospective studies. There were neither prospective nor retrospective cohort studies evaluating the effect of vitrectomy timing on visual outcomes. Characteristics of studies can be found in Table 1.

Table 1.

Characteristics of studies.

Study ID	Country	Number of participants included
Alkuraya 2009 21	Saudi Arabia	1
Mezad-Koursh 2010 22	Israel	7
Emoto 2010 23	Japan	2
Cavalcante 2010 24	United States	1
Flaxel 2011 25	United States	1
Mithal 2013 26	India	7
Englander 2013 27	United States	1
Chaudhary 2013 28	United States	13
Saffra 2014 29	United States	1
Murad-Kejbou 2014 30	United States	2
Thomas 2015 31	United States	1
Nakashizuka 2015 32	Japan	1
Kanchanaranya 2015 33	Thailand	2
Bhavsar 2015 34	United States	1
Singh 2017 35	United States	1
Salceanu 2017 36	United Kingdom	1
Hayes 2017 37	Australia	1
Gensure 2017 38	United States	1
Weinstein 2022 39	United States	1
Romero-Aroca 2012a,40	Spain	3
Irigoyen 2012a,41	United Kingdom	2
Kessner 2014a,42	Israel	7
Simonett 2019a,43	United States	12
Peden 2019a,44	United States	12
Peng 2021a,45	Taiwan	4

^aStudy authors independently provided additional patient-level data not included in original published articles.

Quality assessment

Data for 10 cases were extracted from case reports, and 76 were from case series. All case reports and series met full criteria for quality using the JBI Critical Appraisal tool. The grading of individual studies can be found in Supplemental Material 2.

Immediate versus early versus late PPV in the treatment for endophthalmitis

From the 25 studies included, there were 86 eyes of 86 patients who received treatment for endophthalmitis following intravitreal injection of anti-VEGF agents. There were 37 (43%) patients treated with immediate vitrectomy, 25 (29%) patients with early vitrectomy, and 24 (28%) with late vitrectomy. Demographic data, indication for anti-VEGF, anti-VEGF used, and time frame between anti-VEGF and diagnosis of endophthalmitis can be found in Table 2. There were no remarkable differences between age and gender among the treatment groups. The most common indication for anti-VEGF injection was exudative age-related macular degeneration (wet AMD). The most common anti-VEGF medications used were ranibizumab and bevacizumab. Median time frame between injection and diagnosis of endophthalmitis was 4, 3, and 3 days for the immediate, early, and late vitrectomy groups, respectively (p = 0.610). Culture data was obtained with positive growth of various species for 66 cases (76.7%), and negative growth for 7 cases (8.1%). Results were not explicitly reported per case for the remaining 13 cases (15.1%).

Table 2.

Demographics and characteristics of anti-VEGF injection.

Group	Total	Immediate	Early	Late	p-Value
N	86	37	25	24
Age					0.715
N reported	55 (64%)	22 (59%)	17 (68%)	16 (67%)
Mean (SD)	71 (14)	70 (151)	73 (15)	70 (14)
Range	35–96	35–92	50–96	50–90
Gender					0.055
N reported	43 (50%)	19 (51%)	12 (48%)	12 (50%)
N female (%)	27 (63%)	11 (58%)	6 (50%)	10 (83%)
Indication for anti-VEGF
N reported	69 (80%)	30 (81%)	20 (80%)	20 (83%)
Wet AMD	56	24	18	14
Diabetic retinopathy	7	2	1	4
Retinal vein occlusion	5	3	1	1
Other	2	1	0	1
Anti-VEGF
N reported	84 (98%)	36 (97%)	24 (96%)	24 (100%)
Aflibercept	9	3	3	3
Ranibizumab	31	16	8	7
Bevacizumab	42	16	13	13
Pegaptanib	2	1	0	1
Days between anti-VEGF injection and diagnosis					0.61
Mean (SD)	4 (0.4)	4.4 (2.8)	3.8 (4)	3.6 (0.7)
Range	1–21	1–15	1–21	1–18
Vision at endophthalmitis presentation
>CF		3 (8%)	2 (8%)	6 (25%)	0.887
⩽CF, >LP		28 (76%)	19 (76%)	18 (25%)	0.997
<LP		6 (16%)	4 (16%)	0	0.114

AMD, age-related macular degeneration; Anti-VEGF, vascular endothelial growth factor; CF, count fingers; LP, light perception; SD, standard deviation.

Mean follow-up intervals between diagnosis of endophthalmitis and measurement of final VA were 13, 9, and 10 months (p = 0.598).

Baseline VA prior to causative anti-VEGF injection was available for 30 (81%), 18 (72%), and 16 (67%) patients, respectively. Patient characteristics were similar between patients with and without documented baseline VA.

Across treatment groups, we observed statistical differences in final VA (p < 0.001; Table 3). Significance was likewise observed in the difference and magnitude of change between diagnosis and final VA (p = 0.003, <0.001).

Table 3.

Visual outcomes of treatment groups (logMAR).

Treatment group	Total	Immediate	Early	Late	p-Value
N	86	37	25	24
	Mean (SD)
Baseline	0.7 (0.6)	0.6 (0.5)	1.0 (0.7)	0.5 (0.5)	0.123
Diagnosis	2.2 (0.5)	2.3 (0.5)	2.2 (0.4)	2.0 (0.6)	0.298
Final	1.2 (0.9)	1.0 (0.9)	1.8 (0.9)	0.8 (0.8)	<0.001
Months from diagnosis to final	11.3 (14.5)	13.1 (17)	9.5 (14.5)	10.3 (9.9)	0.598
Difference between
Baseline to diagnosis	1.5 (0.7)	1.6 (0.6)	1.3 (0.7)	1.5 (0.7)	0.238
Baseline to final	0.4 (0.8)	0.4 (0.6)	0.7 (1.0)	0.1 (0.8)	0.088
Diagnosis to final	−1.0 (0.9)	−1.2 (0.8)	−0.5 (1.0)	−1.2 (0.8)	0.003
Magnitude of change between
Baseline to diagnosis	557% (642%)	710% (745%)	302% (388%)	540% (594%)	0.111
Baseline to final	89% (160%)	84% (114%)	114% (168%)	71% (223%)	0.737
Diagnosis to final	−44% (44%)	−56% (35%)	−17.4% (49%)	−55% (39%)	<0.001

SD, standard deviation.

Significant p-values are bolded.

Across groups, we did not observe statistically significant differences and magnitude of change between baseline and diagnosis VA as well as baseline and final VA (Table 3).

The number of patients whose final VA improved from VA at diagnosis was 32 (86%), 12 (48%), and 21 (88%; p = 0.00046). For all patients whose final VA improved from VA at diagnosis, VA improved by at least two lines (0.2 logMAR) or greater. The number of patients whose final VA returned to baseline VA was 11 (37%), 6 (33%), and 10 (62%; p = 0.166).

Of the 10 patients with VA of LP at diagnosis, final VA improved to 20/200 (0.7 logMAR) or better in 1 of the 6 patients who received immediate vitrectomy, and 1 of the 4 patients who received early vitrectomy. Of the remaining eight patients, two patients received immediate vitrectomy and one patient who received early vitrectomy had VA that did not improve from diagnosis. Visual outcomes may be found in Table 3 (Graphs 1–3).

Graph 1.

Visual acuity by case.

Graph 2.

Difference in visual acuity by case.

Graph 3.

Magnitude of change of visual acuity by case.

Magnitude of change from baseline to diagnosis was calculated as (diagnosis − baseline)/(baseline), baseline to final as (final − baseline)/(baseline), and diagnosis to final as (final − diagnosis)/(diagnosis).

Discussion

We found significant differences in final VA across treatment groups. The significance persisted in the magnitude of change analysis between diagnosis and final VA.

Overall, the mean diagnosis and final VA (2.2 and 1.2 logMAR; approximately CF and 20/300) in our study were comparable to that found in a previous systematic literature review evaluating PPV and TAI as initial treatments for endophthalmitis following anti-VEGF injection (1.8–2.2, and 1.2–1.5 logMAR). ¹³

We observed that the immediate and late vitrectomy groups had better median final VA than the early vitrectomy group. While we found no significance between treatment groups regarding improvement from baseline to final VA, the immediate and late vitrectomy groups had greater improvement from diagnosis to final VA. Even when analyzing the magnitude of change, the differences between final VA and improvement from diagnosis to final VA remained significant.

While statistically insignificant, the observed differences in baseline VA across groups may partially contribute to observed visual outcomes. Furthermore, while differences in time between diagnosis of endophthalmitis and final VA measurements were statistically insignificant, we observed shorter follow-up intervals in the early vitrectomy group as compared to the immediate and late vitrectomy groups, which may contribute to the difference in final visual outcomes. The lack of standardized follow-up data limits direct comparison between the visual outcomes of the treatment groups.

Several prior retrospective studies have not specifically analyzed nor found clear consensus regarding the association between early PPV and improved visual outcomes in cases of endophthalmitis. However, studies lack standardization for the definition of vitrectomy timing. “Immediate,” “early,” and “late” PPV have each been considered with various ranges of overlapping timeframes and may not be distinct from timing of PPV as “initial” or “subsequent” intervention.^{8,11,15,46–54} “Initial” or “early” vitrectomy as occurring within the same day of presentation or within multiple days has important implications for clinical practice and management of endophthalmitis. While our study indicates that for our included cases, patients receiving immediate vitrectomy had improved outcomes than those with early vitrectomy, more data is necessary to inform clinical practice and decision-making regarding the immediacy of vitrectomy upon presentation. The development of standard definitions of vitrectomy timing should be considered to aid further research and analysis of treatments.

The theoretical benefits of early vitrectomy include increased volume of sample for culture and microbiological analysis; removal of bacteria, toxins, and material which may impede distribution of intravitreal antibiotics; and clearance of vitreous opacities and membranes. ⁸

The Endophthalmitis post-intravitreal injection clinical trial (EPIIC) is an ongoing clinical trial that aims to evaluate and inform specific management for this disease entity. ⁵⁵ Until the trial findings are published, one may rely on clinical judgment regarding the timing of vitrectomy.

Januschowski et al. found that vitrectomy within 6 h of presentation for postinjection endophthalmitis resulted in improved visual outcomes, though this study is limited by a small sample size of 30 cases as well as inclusion of endophthalmitis following injection of non-anti-VEGF therapies. ¹⁵

The retrospective study by Chaudhary et al. evaluating 23 patients of postinjection endophthalmitis observed that 90% of cases regained baseline VA with initial treatment of TAI only, as compared to 46% that underwent subsequent PPV ranging from 1 to 21 days thereafter. ²⁸ The authors recommended consideration of PPV for patients with worsening clinical course following initial TAI. ²⁸ This study was limited by small sample size and nonspecific analysis of vitrectomy timing.

A meta-analysis by Issa et al. found no significant difference in the visual outcomes between groups that received TAI only versus vitrectomy as the initial treatment. ¹³ However, the study is limited by low-quality evidence and unclear timing of treatment.

For current clinical practice, the 2024 American Society of Retinal Specialists (ASRS) Preferences and Trends Survey reports that for management of a patient with endophthalmitis 48 h after anti-VEGF injection with a visual acuity of HM, 71% of U.S. physicians would perform TAI only without initial vitrectomy, and fewer would perform TAI with subsequent vitrectomy within 6 h (4.6%), within 1–2 days (3.9%), and within 1 week (5.3%). ⁵⁶ This stands in contrast with international respondents for whom only 19.4% would perform TAI only, and more would perform initial vitrectomy within 6 h with or without TAI (53.8%), within 1–2 days (21.2%), and within 1 week (2.6%). ⁵⁶

Study limitations

Due to the restraints of inclusion criteria, this study was limited by the reliance on case reports and case series with associated limitations including selection and reporting biases. We were unable to complete a meta-analysis with the studies available for review. While we implemented a systematic selection protocol, we may have excluded studies meeting inclusion criteria for our review.

The study was limited to the information presented in the included studies and that elicited from authors, if emailed. As a result, we were unable to further characterize nuances related to disease presentation and clinical judgment.

Variations in causality have important implications as they may have different disease profiles with subsequent variations in management.^57,58 Distinguishing between sterile, noninfectious, and infectious endophthalmitis may be challenging. Generally, presenting signs and symptoms of infectious endophthalmitis may be more severe, consisting of more pain, visual disturbance, and greater evidence of inflammation, including hypopyon. ⁵⁹ As the potential for severe, permanent vision loss is greater with infectious etiology, one should clinically have a low threshold for obtaining cultures and administering intravitreal antibiotics. ⁵⁸

While the majority of included cases had positive culture growth with flora isolated via vitreous tap and culture, our study included cases for which cultures were negative or not explicit for patients. As such, our group analysis was unable to completely distinguish between sterile and infectious endophthalmitis, as well as infectious endophthalmitis caused by different pathogens. As the practical intent of our study was to evaluate the effect of PPV timing for patients presenting with signs and findings suspicious of endophthalmitis, initial treatment and decision-making regarding immediate versus early vitrectomy may be significantly limited by laboratory processing of samples, which may take days. In especially virulent cases, delay of vitrectomy may lead to devastating outcomes.

Due to the descriptive heterogeneity in overall treatment courses in the included studies, we only included the timing of PPV in our analyses and did not consider further interventions the patients may have received before the final VA was documented, including repeat PPV or injection of antibiotics. Clinically, the timing and use of intravitreal antibiotics may be important in disease course and visual outcomes. Furthermore, timing of follow-up at which final VA was obtained was not standardized across reported literature, which has implications for evaluation of treatment effect.

While we observed statistically significant findings, our analysis was limited by the small sample size in each of the three treatment groups. Our findings may not be generalizable and serve to summarize the scarce data available regarding the timing of vitrectomy. Performing randomized clinical trials on endophthalmitis is challenging given the relative rarity of the disease. We recommend more detailed reporting regarding the timing and series of TAI and PPV in future case reports and retrospective studies to aid the further evaluation of vitrectomy timing and visual outcomes.

Conclusion

Our results show that there may be an association between time to vitrectomy and visual outcomes, with the immediate and late treatment groups demonstrating better visual outcomes than the early group. Overall, most final visual outcomes improved from vision at diagnosis. Timing of vitrectomy as “immediate,” “early,” and “late” should be standardized with further clarifications of the intervention as initial or subsequent treatments.