Introduction

Intravitreal injections with anti-VEGF agents have drastically improved visual outcomes in patients with various exudative retinal diseases [1]. The use of such injections is increasing globally, in part due to higher prevalence of retinal diseases in aging populations, which greatly increases the number of patients needing therapy [2, 3]. Most patients need regular injections, typically every 4 to 16 months, depending on the type of anti-VEGF and the patient's response to the treatment [4]. Patients with age-related macular degeneration (AMD) report an average total time of 12 h per visit including post-appointment recovery, and caregivers need to take time away from work to accompany patients for more than 20% of care visits [5]. Intuitively, patients who need bilateral intravitreal injection treatment can halve the personal cost and time spent for visits by receiving same-session bilateral injections compared to injections given in each eye on different dates. It is therefore not surprising that more than 90% of patients express a strong preference for same-session bilateral injections over separate, unilateral injection sessions [6]. Third parties benefit accordingly, as salary or leisure time opportunity cost of caregivers are reduced, and potentially reimbursed transportation costs are halved. Moreover, bilateral injection treatment decreases the patient turnover in clinics, which reduces the administrative complexity and enhances the clinical efficiency of treatment centers. Less contacts with health care services have also been a priority during the COVID-19 pandemic.

Serious complications of anti-VEGF injection therapy are generally rare, with the most feared ocular adverse event being endophthalmitis, occurring less than once in 3500 cases [7]. Nevertheless, in bilateral same-session injection therapy, a worst-case scenario is that patients get blind due to severe adverse events occurring in both eyes simultaneously [8], which has been described in singular case reports due to endophthalmitis [9].

In order to reliably investigate the risk of such rare events, very large study populations are needed [8]. For this reason, we have performed a systematic review of the literature to evaluate the risk of endophthalmitis after same-session bilateral intravitreal anti-VEGF injection therapy.

Materials and methods

Study design

This study was a systematic review designed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA). The recommendations of the Cochrane Handbook [10] were followed. According to Danish law, institutional review board approval is not required for such studies. The study protocol was prospectively registered in the PROSPERO database (registration ID: CRD42023405319).

Eligibility criteria

Studies were considered when fulfilling the following criteria:

  • Population: Patients ≥ 18 years of age, being treated with intravitreal injection therapy. No further restrictions on the definition of study participants were enforced.

  • Exposure: Studies were considered in which any kind of anti-VEGF therapy was administered bilaterally in the same setting. If studies did not provide clear statements regarding the timing and setting of injections, it was assumed that studies investigating ‘same-day,’ ‘same-setting,’ or ‘bilateral’ injection therapy had analyzed bilateral injections administered in immediate succession during the same patient sitting. Studies that included intravitreal injections with corticosteroids or antibiotics were excluded due to different drug properties and the potential of dissimilar underlying mechanisms of the diseases being treated, as compared to those treated with anti-VEGF. We did not restrict to any practical aspects of the intravitreal injection, such as the setting (e.g., operating theater or office), the personnel (e.g., doctor or nurse), the device (e.g., prefilled syringes, injection assisting devices, or gauge-size), the underlying retinal condition, or the injected agent (ranibizumab, aflibercept, brolucizumab, bevacizumab, faricimab, or conbercept).

  • Outcome: The aim was to investigate the incidence of post-injection endophthalmitis defined as infectious endophthalmitis. It was assumed that studies investigating or mentioning endophthalmitis without further details were referring to infectious and not sterile endophthalmitis.

  • Study types: All prospective and retrospective studies were eligible for inclusion, regardless of any of the following study designs: randomized controlled trials, observational studies, case–control studies, cohort studies, or cross-sectional studies. We did not consider single case studies or case series.

Information sources and search strategy

One trained author (Y.S.) searched the literature databases PubMed, EMBASE, Cochrane Central, Web of Science Core Collection, BIOSIS Previews, Current Contents Connect, Data Citation Index, Derwent Innovations Index, KCI-Korean Journal Database, SciELO Citation Index, Cumulative Index to Nursing and Allied Health Literature and Zoological Record. No date restrictions were enforced. Studies were considered if disseminated in English or German. Details of the search strategies across literature databases were specified and documented in relation to the search (Supplementary file 1). The search took place on February 26th, 2023. References of studies eligible for inclusion were screened for additional relevant studies.

Study selection, data collection, and risk of bias within studies

One author (J.B.) examined titles and abstracts from the literature search and removed duplicates and obviously irrelevant reports. Two authors (M.S. and J.B.) then independently examined full text of remaining references for eligibility and reviewed references from these studies for any additional relevant studies. Afterward, consensus was attempted within the study group and in case of further disagreement, a third author (Y.S.) was invited to discuss and to reach a final consensus. Data regarding study design, characteristics, methods, and results were extracted from eligible studies using extraction forms.

To assess the quality of the studies, the Newcastle–Ottawa Quality Assessment Scale (NOS) for Cohort Studies toolkit was used, which evaluates categories within three domains: selection, comparability, and outcome. Categories within selection are representativeness of the exposed cohort, selection of the non-exposed cohort, ascertainment of exposure, and demonstration that outcome of interest was not present at start of study. For comparability, one category evaluated is comparability of cohorts on the basis of the design or analysis. Categories within outcome are assessment of outcome, whether follow-up long enough for outcomes to occur, and adequacy of follow-up of cohorts. 0–2 points are given comparability criteria while 0–1 points are given for other criteria. The quality score of 0–9 is a summary of the number of points across all categories within each study. The risk of bias assessment was conducted by two authors (O.K. and J.B.) independently. In case of disagreement, a third author (Y.S.) was invited to reach consensus.

Synthesis of results

All studies were reviewed qualitatively in text and tables. Total single injections, total bilateral injection sessions, and mean injections per patient were recorded from study publications if stated or was otherwise calculated if the data was available. It was assumed that one bilateral injection equaled two single injections if no other information was provided. Incidence of endophthalmitis was calculated from studies that presented either positive or negative statements on the occurrences of endophthalmitis, or if it was otherwise obvious that the investigators had been attentive to endophthalmitis but found no cases.

Results

Study selection

The literature search found 3,324 titles and abstracts, of which 1,616 duplicates and 1,676 obviously irrelevant reports were discarded. Thirty-two remaining publications were reviewed in full-text, 15 of which were discarded for reasons: not original data (n = 5), reporting on injections of both anti-VEGF and triamcinolone injections (n = 3), bilateral injections not performed during the same session (n = 3), case report (n = 1), published in neither English nor German (n = 2), or pre-publication conference abstract of an already included study (n = 1). Consequentially, 17 studies were found eligible for inclusion (Fig. 1), which summarized data of 138,478 intravitreal anti-VEGF injections (69,239 bilateral injections sessions) [6, 8, 11,12,13,14,15,16,17,18,19,20,21,22,23,24,25]. In one study, the exact number of injections was not disclosed, but 87 patients had bilateral same-session injections during a 1-month long study period, from which we interpret that 174 injections (87 bilateral injection sessions) were given [25]. No additional studies were found from screening references of included studies.

Fig. 1
figure 1

PRISMA 2020 flow diagram for systematic reviews which includes searches of databases, registers, and other sources. WSCC Web of science core collection, BIOSIS BIOSIS previews, CCC Current contents connect, DCI Data citation index, DII Derwent innovations index, KCI KCI-Korean journal database, SciELO SciELO citation index, ZR Zoological record

Full size image

Study characteristics

Fifteen studies were of retrospective nature, with all but one [20] being clearly defined as single-center and register-based, while two were prospective, randomized clinical trials [15, 25]. Eight studies originated from the USA [6, 8, 13, 15, 18, 19, 21, 24] and the rest from Lithuania [20], the UK [22], Jordan [23], France [25], Nigeria [11], Spain [12], Canada [14], South Korea [16], and Pakistan [17]. The studies were published between 2009 and 2022 examining injections administered between 2006 and 2019. (Table 1).

Table 1 Study demographics
Full size table

The highest number of injections came from the study by Borkar et al. with 101,932 total single injections (50,966 bilateral injection sessions) followed by Jeeva et al. with 15,338 injections. Five studies contributed between 1,000 and 10,000 injections [12, 14, 16, 18, 19] and the remaining studies included less than one thousand injections each [6, 11, 13, 15, 20,21,22,23,24,25]. In total, at least 7579 patients were included (two studies did not disclose the number of patients [12, 17]). Among studies that disclosed the number of included patients, Borkar et al. had the largest study with 5,890 subjects, followed by Lima et al. and Jang et al. with 367 and 323 patients, respectively. Five studies included between 100 and 300 patients [6, 14, 19] and the remaining studies included less than 100 patients each.

The age of participants was disclosed in 10 studies [6, 8, 11, 13,14,15,16, 19, 21, 22]. The mean age of patients ranged from 55.7 to 82.5 years across these studies.

Neovascular age-related macular degeneration was the sole indication for injection therapy in four studies [6, 19, 20, 22] and the primary indication for treatment in six studies [8, 14, 16, 18, 21, 25], while diabetic macular edema was the primary indication in four studies [11, 13, 15, 23]. Three studies did not disclose either the indications or the number of patients with certain indications for treatment [12, 17, 24] (Table 2).

Table 2 Therapy details and study outcomes
Full size table

Details of injections

Seven studies investigated injections with bevacizumab, ranibizumab, and aflibercept [8, 13,14,15,16,17, 24]. Three studies examined ranibizumab and bevacizumab injections [6, 18, 19, 23] and one study included ranibizumab and aflibercept [12]. The three remaining studies were with bevacizumab only [11, 20] or ranibizumab only [21, 22] or without disclosure of the type of anti-VEGF agent [25]. Excluding the three studies that did not reveal the relative number of injected agents [13, 19, 25], 52.2% of total injections across all studies were ranibizumab, 28.6% were aflibercept and 25.6% bevacizumab.

Injections were given by ophthalmologists [11,12,13, 19], consultants [11, 22, 23], an unspecified treating physician [21], qualified or trained ophthalmology residents [11, 23], physicians in training under direct supervision of retina specialists [24], vitreoretinal fellows or specialists [8, 14] or a vitreoretinal fellowship trained retinal specialist [8], while 6 studies provided no information regarding proceduralists (Table 3).

Table 3 Details of injection procedures
Full size table

Injection technique was specified in 14 studies. In 7 studies it was explicitly stated that injections were delivered according to local or national standard protocols [13, 17,18,19, 22, 24]. The injection setting was provided by 8 studies and was office-based [8, 14, 18], a dedicated outpatient or treatment room [16, 22], ophthalmic operating room [17, 23], or a controlled ambient surgery cabin [12]. All but two studies specified that a new set of sterile equipment was used for each eye, but exact details varied. In one study, the same bottle of anesthetic and disinfectant was used for fellow eyes [8]. In two studies, masks and gloves were not used by physicians, but silence was maintained during injections [8, 12]. Topical anesthesia included topical drops and/or application of cotton swabs soaked in an anesthetic agent administered once or several times and/or subconjunctival injections. In studies that disclosed the information, lid speculum was always used in 8 studies, never used in 2 studies [8, 15] and mostly used in one study [14]. Povidone-iodine was the ocular disinfectant used in all but one study, which was a clinical trial on aqueous chlorhexidine and povidone-iodine [15]. Disinfection of the periorbital skin and eyelids was done in 6 studies [11, 16, 17, 19, 21, 23]. Across the studies, 30-, 31- or 32-gauge needles were used, and anti-VEGF injections were drawn from single-use vials or came as prepacked syringes from compounding pharmacies. Only one study stated explicitly that anti-VEGF injections for fellow eyes came from different batch numbers [17]. Injection sites were mostly specified as 3–4 mm posterior to the limbus, and the distance was measured by caliper in four studies [6, 11, 17, 21]. After injection, a sterile cotton tip was applied to the injection site in four studies [6, 11, 14, 21], and in most studies topical antibiotics were prescribed, ranging from fluoroquinolone [6, 11, 12, 16,17,18,19, 23] or Polymyxin B and Trimethoprim [18] to chloramphenicol [22]. (Table 3).

Risk of bias within individual studies

Newcastle–Ottawa Scale study quality scores were generally mediocre, mainly due to the lack of control groups (9 studies; Selection item #2) and inadequate or no comparability between intervention and control groups (15 studies; Comparability item). Lack of adequate follow-up time for the outcome of endophthalmitis to occur (Outcome item #2) and no statements on whether loss to follow-up had occurred (Outcome item #3) also negatively impacted quality scores for one-third of studies. The highest quality score was found for Mahajan et al. [6] with a quality score of 8, while Lima et al. [18], Cimbalas et al. [20], and Okoye et al. [11] were given the lowest score of 4 (Table 4). We choose to grade studies equally regarding Selection item #4 (‘demonstration that outcome of interest was not present at start of study’) and give all studies a point, due to the following reasons: It is unlikely for patients to have had endophthalmitis before injections, and we consider it common practice to cancel scheduled injection treatment or reassess patients if they present with obvious clinical signs of endophthalmitis.

Table 4 Study quality of included studies using the Newcastle–Ottawa Quality Assessment Scale for Cohort Studies
Full size table

Incidence of endophthalmitis following bilateral injections

A total of 33 cases of unilateral endophthalmitis were reported after 138,478 injections (69,239 bilateral injections sessions). No cases of bilateral endophthalmitis were seen. The rates of endophthalmitis per single injection in patients receiving bilateral same-session injections ranged from 0.000 to 0.526% (0.526% in the study by Shah et al., 0.146% in Abu-Yaghi et al., 0.065% in Lima et al.; 0.027% in Borkar et al., 0.010% in Juncal et al., 0.007% in Jeeva et al. and zero in the remaining 11 studies).

There were 8 studies that compared an intervention group of patients treated with same-session bilateral injections with a control group of patients treated with unilateral injections, either in one eye only or in both eyes on different dates [6, 12, 17, 20, 21, 23,24,25]. In total, intervention groups contributed 19,475 same-session bilateral injections, after which two cases of unilateral endophthalmitis had occurred; control groups totaled 31,109 unilateral injections with 7 cases of post-injection endophthalmitis. No cases of bilateral endophthalmitis were described.

Discussion

We have provided a review of the current literature on the risk of endophthalmitis after bilateral same-session intravitreal anti-VEGF injections. Previous meta-analyses on the risk of endophthalmitis after intravitreal injections without specification of treatment laterality have agreed on endophthalmitis rates of 0.05–0.06% [26, 27]. Approximately half this rate was found in the presently largest study on bilateral same-session injections by Borkar et al., which had a sound methodology and contributed to 74% of injections included in our review [8]. Although a high risk of bias should be acknowledged in such direct comparisons without meta-analysis, it appears that the risk of endophthalmitis after bilateral same-session injections is at a clinically reasonable level.

No cases of bilateral endophthalmitis were reported in any of the studies included in this review. This is reassuring, as endophthalmitis is perceived as the most serious adverse event in intravitreal anti-VEGF injection therapy, after which permanent vision loss is commonly seen, and evisceration may be needed in severe cases [28].

We only considered infectious endophthalmitis in this review, either proven by vitreous cultures [8, 18, 23] or otherwise defined as infectious or presumed infectious [8, 14, 17]. Studies disclosed clear specifics regarding only 4 of the 33 found endophthalmitis cases [14, 18, 23], and group demographics and clinical characteristics were described for 28 additional cases [8]. One study provided limited specifics of endophthalmitis cases, but it was unclear which cases belonged to the groups of unilaterally or bilaterally treated patients [17]. Clear statements of positive vitreous cultures were only present regarding 3 cases, in which of Streptococcus viridans, Staphylococcus epidermidis, and coagulase- negative staphylococcus had been found [18, 23], with antibiotic sensitivity profiles only disclosed for the first two cases [18]. Due to the limited availability of specific microbiological data from these studies, this information is not suitable for preventative purposes.

Several included studies considered cases of sterile inflammation separately from cases of infectious endophthalmitis, heterogeneously defined as sterile vitritis or non-infectious endophthalmitis [11, 19, 22], severe acute intraocular inflammation [18], acute or sterile acute intraocular or ocular inflammation [12, 14, 16], sterile inflammation [6], anterior chamber cell or flare [11, 19], and uveitis [24]. Due to the non-harmonized definitions used, we did not find that a meaningful analysis could be done regarding sterile inflammation.

Despite the lack of large randomized or controlled trials demonstrating safety in bilateral same-session intravitreal injection [25], surveys of intravitreal injection practices among retinal specialists in the USA reveal that the proportion of practitioners that perform bilateral simultaneous injections is high and have remained stable from 2011 to 2019, at approximately 45% [29, 30]. In this systematic review, we found that endophthalmitis is a rare complication following bilateral same-session anti-VEGF injection therapy. Our study found no cases of bilateral endophthalmitis following bilateral same-session injections, which leaves such events to have been described only in rare case reports [9, 31]. Based on a conservative independent risk of endophthalmitis at 0.09% per injection, the risk of bilateral blindness from two independent, sterile procedures should be only 1 in 1.2 million injections [31]. The real-life risk is likely higher, as same-session bilateral procedures will never have a completely detached risk of endophthalmitis, as the treatment room, proceduralists, patient and post-injection patient environment and behavior is the same [31]. In any case, it seems that half of practitioners agree that the convenience and cost benefits of same-session bilateral injections outweigh the clinical risks of such extremely rare, serious patient complications [29, 30]. A prerequisite for acceptance is that injections are treated as separate, sterile procedures, and conducted according to current evidence for endophthalmitis prevention [31].

Other potentially vision-threatening ocular adverse events to anti-VEGF injection therapy include immune-mediated sterile inflammatory reactions, acute increase in intraocular pressure, and retinal detachment [32]. Risk of systemic adverse events, including death, after bilateral anti-VEGF injection therapy has also been a topic of concern [23, 33]; however, evidence so far does not provide a clear direction on this matter. Further studies may be warranted.

Limitations need to be acknowledged regarding this review and its findings. First, our study found that the number of studies and thus total number of bilateral same-session injections published in the literature are limited, and a meta-analysis was not mandated. Findings should therefore not be interpreted as conclusive. Second, lack of homogenous and detailed data across studies, including regarding anti-VEGF drugs, treatment settings and injection providers, did not allow stratified analyses. Finally, since most included studies were retrospective, various sources of bias could have influenced their findings, which may explain the heterogeneity in findings observed across included studies. However, considering the low incidence of endophthalmitis found in this review, even a pragmatic randomized clinical trial design would need a disproportionally high number of participants and would be challenging to carry out in practice.

Strengths of this study include the literature search strategy including 12 databases, which allowed as many studies as possible to be considered in a comprehensive literature review, as well as adherence to best practice methodology in the study design.

Conclusion

In conclusion, the current literature suggests that the rate of endophthalmitis following bilateral same-session intravitreal anti-VEGF injections is at a clinically acceptable, low level. We suggest that clinicians can consider same-day treatment of both eyes of patients in need of bilateral intravitreal anti-VEGF injection therapy without compromising safety. With an increasing demand for intravitreal therapy, same-day bilateral treatment may hold the potential to improve clinical efficiency and patient satisfaction.