Aims To describe and compare the causative organisms, clinical features and visual outcomes of endophthalmitis following intravitreal injection (IVI) to endophthalmitis following cataract surgery.
Methods Patient population and setting: A retrospective case series of patients with acute endophthalmitis following either cataract surgery or IVI presenting to a tertiary referral centre—Sydney Eye Hospital—between 2007 and 2010. Main outcome measures: (1) identification of the causative organism; (2) time to presentation; (3) odds of improvement in visual acuity (VA) following treatment; (4) odds of final VA of counting fingers (CF) or less and (5) odds of enucleation.
Results Of the 101 patients in our study, 48 had preceding cataract surgery and 53 had preceding IVI. There was an increased incidence of Streptococcus spp. endophthalmitis in post-IVI cases (24.53% vs 6.25%; OR 5.85; p=0.022). Endophthalmitis following IVI had increased likelihood of a final VA of CF or less (OR=6.0; p<0.01), decreased likelihood of any improvement in acuity following treatment (OR=0.13; p<0.01) and an increased likelihood of presenting within a week of the procedure (OR=3.93; p<0.01). Endophthalmitis caused by Streptococcus spp. was associated with increased likelihood of a final VA of CF or less (OR=10.2; p<0.01), decreased likelihood of any improvement in acuity following treatment (OR=0.06; p<0.01) and increased likelihood of enucleation (OR=17.11; p<0.01).
Conclusions Endophthalmitis following IVI is associated with an increased incidence of Streptococcus spp. infection, earlier presentation and poorer visual outcomes when compared with endophthalmitis following cataract surgery.
- cataract surgery
- intravitreal therapy
- visual perception
- colour vision
Statistics from Altmetric.com
- cataract surgery
- intravitreal therapy
- visual perception
- colour vision
Endophthalmitis is an infrequent—though visually devastating—complication of intraocular surgery. The incidence of acute (within 6 weeks of surgery) endophthalmitis following cataract extraction has been estimated to range from 0.04%1 to 0.3%.2 The most commonly cultured organism from vitreous samples is coagulase-negative Staphylococcus spp.1 ,3–5
Over recent years, developments in ocular pharmacotherapeutics has seen a rapid increase in the number of intravitreal injections (IVIs) performed, especially of anti-vascular endothelial growth factor agents for exudative age-related macular degeneration. The incidence of endophthalmitis following IVI of anti-vascular endothelial growth factor agents (pegaptanib, ranubizumab or bevacizumab) has been reported to range from 0.019%6 to 0.54%.7 A recent meta-analysis drawing together the available small case series suggests that while Staphylococcus spp. are still the most commonly cultured causative organisms, Streptococcus spp. infection occurs three times more commonly when compared with other forms of intraocular surgery.8 This observation is of significant concern, both because of the reported poor outcomes associated with Streptococcus spp. endophthalmitis,9 ,10 and because the proposed mechanism of inoculation is via droplet contamination/aerosolisation of saliva (Streptococcus spp. forms approximately 40% of the salivary flora)11 contaminating the operative field. If the spectrum of causative organisms in endophthalmitis following IVI differs to that following intraocular surgery, then the visual outcomes and optimum management may also be significantly different.
To date, the only study to find a statistically significant increase in the proportion of endophthalmitis cases caused by Streptococcus spp. infection in post-IVI endophthalmitis (compared with post-surgical endophthalmitis) is a meta-analysis of cases separated in time and location.8 Subsequently published case series only hint at the possibility that post-IVI endophthalmitis is associated with an increased incidence of Streptococcus spp. infection.10 ,12 ,13 Indeed, the only direct comparison to date (of a cohort of 88 post-surgical and 21 post-IVI cases) by Chen and colleagues12 failed to show a statistically significant increased incidence of Streptococcus spp. infection in post-IVI endophthalmitis. Furthermore, no study has compared directly the time to presentation or clinical outcomes of post-IVI with post-surgical endophthalmitis.
The aim of our study was to directly compare the clinical characteristics, spectrum of causative organisms and outcomes of acute endophthalmitis following IVI with that following cataract surgery.
The institutional review board approved this retrospective, non-randomised comparative chart review performed at Sydney Eye Hospital, a tertiary referral centre for the state of New South Wales, Australia. The charts of all cases of endophthalmitis—regardless of where the inciting procedure was performed—presenting to Sydney Eye Hospital during time period 2007–2010 were reviewed. We included only those cases with a diagnosis of microbial endophthalmitis, or presumed microbial endophthalmitis, that occurred within 6 weeks of either an IVI or cataract surgery.
Case-related data collected included presentation and last follow up of Snellen visual acuity (VA), number of days from inciting procedure until presentation, whether or not a trans-pars plana vitrectomy was performed, and the length of time from presentation until vitrectomy, culture and sensitivity results, use of systemic antibiotics, length of follow-up, development of significant further complications such as retinal detachment or neovascular glaucoma, and whether or not enucleation was ultimately required.
All patients in this study underwent a standard trans-pars plana vitreous ‘tap and inject’ protocol at presentation. A 0.2 ml vitreous sample was taken for microbiological assessment following which 1 mg of vancomycin in 0.1 ml and 2.25 mg of ceftazidime in 0.1 ml were injected intravitreally. All cases had symptoms consistent with endophthalmitis, which included increasing ocular pain and loss of vision. All cases had documented signs of endophthalmitis including anterior chamber hypopyon and fibrin, loss of red reflex secondary to vitreous opacification and vitreous opacities and membranes present on B-scan ultrasound. One or more vitreoretinal specialists confirmed the diagnosis of presumed microbial endophthalmitis in all cases. The decision to use systemic antibiotics or to perform a vitrectomy was at the discretion of the managing ophthalmologist.
Clinical characteristics of post-cataract surgery and post-IVI patients and microbiological factors associated with poor outcomes—in terms of final acuity and need for enucleation—were compared by computing odds and ORs (including 95% CIs). VAs were converted to logMAR values, as described by Ferris et al.14 A VA of hand movements was assigned a logMAR of +3.0, and counting fingers (CF) VA was assigned a logMAR of +2.0, according to methods published by Holladay.15 Additionally, we assigned an acuity of light perception a logMAR of +4.0. No logMAR value was attached to total blindness (ie, no light perception, NLP); however, such outcomes were ranked below light perception in our analysis ( see below ). Statistical significance was assessed via Fisher's exact test (two-tailed) and further analysis of the interactions between risk factors for poor outcomes and early presentation was performed using three-way log-linear analysis (two-tailed). All test statistics were computed using the VassarStats package.16 Significance levels were adjusted for multiple comparisons using the SDM Project FDR calculator,17 which employs the false discovery method proposed by Benjamini and Hochberg.18
A total of 101 patients met the criteria for inclusion in our study. Of these, 38 were culture negative and 63 were culture positive: the causative organisms in culture-positive cases in each group are summarised in table 1. A total of 48 cases were post-cataract patients while 53 were post-IVI (35 post-injection of ranubizumab; 12 post-injection of bevacuzimab; 6 post-injection of triamcinolone). The average age of post-IVI patients was 73.1 years (95% CI 68.9 to 77.3 years) and the average age of post-cataract surgery patients was 72.7 years (95% CI 69.6 to 75.7 years). Mean time to presentation was 4.5 days (95% CI 3.7 to 5.4 days) for post-IVI endophthalmitis and 7.2 days (95% CI 5.2 to 9.3 days) for post-cataract surgery endophthalmitis. Mean follow up was 35.4 weeks (95% CI 28.7 to 42.0 weeks) for post-IVI patients and 31.3 weeks (95% CI 23.3 to 39.3 weeks) for post-cataract surgery patients. Median final acuity in the post-cataract surgery group was +0.40 logMAR (range +0.10–NLP), while median final acuity in the post-IVI group was +2.0 logMAR (range +0.18–NLP).
We compared various parameters, including presenting acuity, final acuity, change in acuity, culture results from vitreous tap samples, enucleation rates, progression to trans-pars plana vitrectomy, post-operative retinal detachment ratesi and addition of systemic antibiotics in patients suffering from endophthalmitis following either cataract extraction or IVI (table 2). We also examined microbiological factors associated with outcomes of VA ≤CF (table 3), any improvement in VA (table 4) and enucleation (table 5) by specifically comparing the odds of these outcomes in cases of Streptococcus spp. endophthalmitis versus other culture-positive endophthalmitis and Staphylococcus epidermidis endophthalmitis versus other culture-positive endophthalmitis (S epidermidis and Streptococcal spp. were the two most commonly encountered species overall).
We found no significant difference between endophthalmitis following cataract surgery and endophthalmitis following IVI in terms of presenting acuity, odds of culture positivity, odds of vitrectomy, odds of systemic antibiotic use, odds of enucleation or odds of retinal detachment (table 2). In terms of outcomes, patients presenting post-IVI had an increased likelihood of having a final acuity of CF or less (OR=6.0; 95% CI 2.26 to 15.93, p<0.01). Conversely, endophthalmitis following IVI was less likely to show any improvement in VA following treatment (OR=0.13; 95% CI 0.04 to 0.39, p<0.01). Those with endophthalmitis following IVI had an increased likelihood of presenting within a week of the procedure (OR=3.93; 95% CI 1.52 to 10.13, p<0.01). Endophthalmitis following IVI was associated with an increased likelihood of culture positivity for Streptococcus spp. compared with endophthalmitis post-cataract extraction (OR 5.85; 95% CI 1.47 to 23.30, p=0.022). Streptococcus spp. endophthalmitis, regardless of the mechanism of inoculation, was associated with worse outcomes in terms of an increased likelihood of a final acuity of ≤CF (OR=10.2; 95% CI 2.02 to 51.1, p<0.01), a decreased likelihood of any improvement in VA (OR=0.06; 95% CI 0.01 to 0.27, p<0.01) and an increased likelihood of enucleation (OR=17.11; 95% CI 3.04 to 96.26, p<0.01). Conversely, S epidermidis endophthalmitis was associated with a significantly lower likelihood of a final VA of ≤CF (OR=0.07; 95% CI 0.02 to 0.27, p<0.01), an increased likelihood of any improvement in VA (OR 43.63; 95% CI 5.18 to 367.72, p<0.01) and a decreased likelihood of enucleation (no cases; p<0.01).
In order to distinguish the relative contribution of culture results (ie, Streptococcus spp. endophthalmitis vs other culture-positive endophthalmitis) and the inciting procedure (ie, cataract surgery or IVI) to poor outcomes and early presentation, we also undertook a three-way log-linear analysis of our data. Specifically, this procedure explored the interactions between inciting procedure, culture result (Streptococcus spp. vs non-Streptococcus spp. infection) and visual outcome or time to presentation. This analysis suggests that Streptococcus spp. endophthalmitis is an independent risk factor for poor visual outcomes (in terms of VA ≤CF; G2=7.76, p=0.039, or any improvement in acuity; G2=8.56, p=0.028) and risk of enucleation (G2=14.72, p<0.01). IVI—once the effect of the increased incidence of Streptococcus spp. was accounted for—was not found to be independently associated with an increased risk of poor visual outcomes (in terms of VA ≤CF; G2=5.54, p=0.09, or any improvement in acuity; G2=5.50, p=0.09) or enucleation (G2=4.00, p=0.18). In terms of early presentations, three-way log linear analysis suggests that Streptococcus spp. infection was not independently associated with presentation within a week of the procedure (G2=0.96, p=0.65) once the effects of the inciting procedure were eliminated. Conversely, IVI was found to be independently associated with early presentation (G2=7.12, p=0.048) after the increased incidence of Streptococcus spp. was accounted for.
Our study is the first to directly demonstrate in a single retrospectively studied cohort a statistically significant increased incidence of Streptococcus spp. endophthalmitis in patients presenting post-IVI when compared with patients presenting post-cataract surgery. Culture positivity for Streptococcus spp. has previously been described to result in poor outcomes9 ,12 and was associated in our group with increased odds of poor final VA (≤CF), decreased odds of any improvement in vision and increased odds of enucleation. Log-linear analysis of the three-way interaction between inciting procedure, culture result and outcomes confirms that Streptococcus spp. infection remains a risk factor for poor visual outcomes (final VA of ≤CF and no improvement in acuity following treatment) and for enucleation once the effect of the inciting procedure is eliminated.
A number of factors other than infection-induced pathology might be expected to account for poorer final acuity in post-IVI endophthalmitis compared with post-cataract surgery endophthalmitis. It is anticipated that the average patient undergoing cataract surgery has a superior visual potential to the average patient undergoing IVIs. The latter patients tend to have some degree of irreversible structural damage from their disease process (eg, age-related macular degeneration, diabetic retinopathy etc); furthermore, such patients may have active disease, and their treatment regime is commonly interrupted during the course of acute infection. Ideally, the analysis of final acuity data would take into account baseline acuity: a specific limitation of our study, therefore, is that such data were not available. However, the post-IVI group was significantly less likely to experience any VA improvement with treatment compared with the postoperative cataract group and were more likely to have a final acuity of ≤CF. Furthermore post-IVI endophthalmitis was not associated with either of these outcomes once the increased frequency of Streptococcus spp. infection was accounted for using a three-way log-linear analysis. This observation—combined with the fact that Streptococcus spp. infection was independently associated with poor outcomes—suggests that the poor outcomes associated with post-IVI endophthalmitis may be attributed to the increased incidence of Streptococcus spp. infection in this group of patients.
The distribution of causative organisms in our study was broadly similar to that described in the meta-analysis of McCannel,8 which suggests that our findings have relevance beyond the confines of southeast Australia. Furthermore, our study has the advantage that it provides a direct comparison of a large group of patients, rather than relying upon statistical analysis of cases separated in time and location. It has previously been noted that the spectrum of micro-organisms cultured from vitreous samples in post-operative endophthalmitis resembles that of the ocular surface flora, which is dominated by coagulase negative Staphylococcus (culture positive in about 76% of patients), with lower frequencies of Staphylococcus aureus (culture positive in about 7%) and Streptococcus spp. (culture positive in about 7%).19 Endophthalmitis following IVI clearly departs from these trends in terms of the distribution of causative organisms, with an over-representation of Streptococcus spp. McCannel8 has proposed that other interventional procedures, such as intra-articular injections and lumbar punctures, may provide useful insights into the mechanism of infection in intravitreal endophthalmitis because these non-ocular invasive procedures may follow similar steps of preparation and may be similarly complicated by Streptococcal spp. infection. It is known that Streptococcus spp. forms an important part of the normal oral flora,20 and it has previously been proposed that Streptococcus spp. infections associated with intra-articular injections and lumbar punctures result from inoculation from oral flora via droplets or aerosolisation.21–25 Furthermore, it has been suggested that wearing of face masks, or refraining from talking or coughing during injections, may decrease the frequency of Streptococcus spp. endophthalmitis.8 Although there are data which suggest that wearing of face masks may decrease the likelihood of bacterial contamination within operating fields,22 ,26 ,27 critics of such studies suggest that face masks themselves may induce bacterial shedding from the skin surface and may impair communication.28 At present, the suggestion that face masks may reduce the incidence of Streptococcus spp. endophthalmitis is a hypothesis that is yet to be directly tested.
We also found that endophthalmitis following IVI presented earlier than endophthalmitis following cataract surgery in our patient group. Log-linear analysis of the three-way interactions between inciting procedure, culture result and time to presentation shows that this difference cannot be attributed to the increased incidence of Streptococcus spp. infection; rather, it appears to be due to other factors associated with IVI. This observation may reflect differences in the ability of inoculated organisms to multiply in the vitreous compared with the anterior chamber: such a hypothesis is supported by animal models of endophthalmitis, which demonstrate that the vitreous cavity offers an environment which is more conducive to bacterial survival and growth than the anterior chamber.29
In summary, endophthalmitis following IVI presents earlier, is associated with significantly worse visual outcomes (increased likelihood of VA ≤CF; decreased likelihood of any improvement in VA following treatment) and is associated with an increased frequency of Streptococcus spp. infection when compared with endophthalmitis following cataract surgery. Analysis suggests that the earlier presentation of post-IVI endophthalmitis in our group of patients cannot be attributed to the increased frequency of Streptococcus spp. infection, but rather is due to other factors associated with the inciting procedure. Conversely, the increased visual morbidity associated with post-IVI endophthalmitis does appear to be related to the increased incidence of Streptococcus spp. infection.
Competing interests None.
Ethics approval Ethics approval was provided by the Sydney Eye Hospital Ethics Committee.
Provenance and peer review Not commissioned; externally peer reviewed.
↵i One patient with retinal detachment also developed neovascular glaucoma—this was the only case of the latter complication.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.