Purpose
To present cases of endophthalmitis following intravitreal injections where povidone-iodine (PI) was not used as part of the surgical preparation.
Design
Retrospective case series.
Methods
All cases of presumed injection-related endophthalmitis presenting to the Massachusetts Eye and Ear Infirmary between June 2008 and November 2014 and Dean McGee Eye Institute between January 2010 and January 2015 were identified. Patients who did not receive PI preparation owing to documented self-reported allergy to iodine, iodine-containing contrast material, or shellfish were identified and their injection histories and clinical courses reviewed.
Results
The combined rate of postinjection endophthalmitis at these 2 centers was 0.019%. Among 42 patients with postinjection endophthalmitis, 5 (11.9%) did not receive PI prophylaxis. The mean number of intravitreal injections without PI before the development of endophthalmitis was 10.6 with a 9.4% rate of endophthalmitis (5 cases per 53 injections). All patients underwent tap-and-inject procedures with vancomycin 1 mg and ceftazidime 2 mg. Two patients did not receive PI at the time of tap and inject; 1 of these patients required subsequent pars plana vitrectomy for worsening clinical course. Cultures were positive in 4 of 5 cases; all positive cultures grew coagulase-negative Staphylococcus . All patients who received subsequent intravitreal injections received PI prophylaxis without allergic reactions, thus demonstrating a lack of true PI allergy.
Conclusions
Avoiding PI owing to self-reported iodine “allergy” risks substantial ocular morbidity. Allergy testing can be pursued per patient request or in rare cases of suspected true PI allergy; however, in cases where delayed treatment would adversely affect visual outcome, the clinician should feel confident that minimal allergic risk exists.
Limiting the risk of postinjection endophthalmitis is an area of considerable practical and academic interest, especially in the era of regular intravitreal injections. Although the incidence of postinjection endophthalmitis is low (0.056% per injection in a recent meta-analysis), the risk to an individual patient is magnified by the recurrent nature of the procedure. Using povidone-iodine (PI) for surgical antisepsis is well established and is the only preoperative measure shown to reduce the risk of endophthalmitis in patients undergoing intraocular surgery. The importance of PI application prior to intravitreal injection has been observed by Nentwich and associates, Bhavsar and Sandler, and Brynskov and associates, the latter of whom described no cases of endophthalmitis after 20 293 injections. Patients with a reported allergy to iodine or iodine-containing agents present a clinical challenge and some clinicians modify the standard ophthalmic preparation to avoid PI because of perceived iodine allergy. This study presents a series of cases of intravitreal injection–related endophthalmitis in patients who underwent modified surgical preparation owing to reported iodine allergy.
Methods
Institutional review board (IRB) approval for a retrospective chart review was obtained from the Massachusetts Eye and Ear Infirmary Human Studies Committee and the University of Oklahoma Health Sciences Center IRB. All cases of presumed intravitreal injection–associated endophthalmitis diagnosed at 2 institutions—Massachusetts Eye and Ear Infirmary (MEEI, Boston, Massachusetts), between June 1, 2008 and November 1, 2014, and Dean McGee Eye Institute (DMEI, Oklahoma City, Oklahoma), between January 1, 2010 and January 1, 2015—were identified using International Classification of Disease (ICD)-9 billing codes. Patients who had received ophthalmic care by outside doctors and subsequently presented to either institution were included, in addition to those who were established patients at the study centers. Patients with documented self-reported allergy to iodine, iodinated contrast material, or shellfish were identified and their charts further reviewed to determine whether PI was instilled as part of the antiseptic protocol prior to the injection that preceded endophthalmitis as well as subsequent tap and inject or pars plana vitrectomy. Additionally, complete procedure history was reviewed, including all intravitreal injections before and after endophthalmitis as well as further demographic and clinical characteristics including age, sex, medication injected, ocular condition being treated, interval between last injection and endophthalmitis presentation, microbiology results, visual acuity (at presentation and last follow-up), and number of injections (before and after endophthalmitis) with and without preinjection PI antisepsis.
Results
Injection protocols at both MEEI and DMEI were reviewed. The procedures were office-based at both institutions. To achieve anesthesia, proparacaine and lidocaine were employed. Topical lidocaine was the norm at DMEI, whereas physicians were divided evenly between topical and subconjunctival lidocaine use at MEEI. Topical PI was applied to the ocular surface; eyelid scrubs were not typically used. Masks and eyelid drapes were not used but speculum, gloves, and caliper use were the norm at both institutions. Topical antibiotics were typically not given after injections; however, 2 patients (Cases 2 and 3) who did not receive PI at the time of their injection received postinjection topical antibiotics, and 1 patient who did receive PI and went on to develop endophthalmitis (at MEEI) also received topical antibiotics. None of the patients with endophthalmitis (with or without PI use at the time of injection) at DMEI received postinjection antibiotics.
There were 30 046 injections performed at MEEI during the study period and 6 of these resulted in endophthalmitis, giving a 0.020% rate of postinjection endophthalmitis. There was a 0.018% rate of postinjection endophthalmitis for patients injected at DMEI (33 699 injections, 6 infections). The combined rate of endophthalmitis was 0.019% at both institutions. There were 21 postinjection endophthalmitis cases from outside doctors who referred the patients to MEEI and 9 cases referred to DMEI; however, because the total number of injections in the patient catchment area is unknown, the regional rate of endophthalmitis outside these institutions could not be calculated. In total, 42 cases of presumed intravitreal injection–related endophthalmitis were identified: 27 cases at MEEI and 15 at DMEI.
Five of the 42 patients (11.9%) were identified as having a history of self-reported prior allergic reaction to iodine-related agents and did not receive PI antisepsis at the time of the injection immediately preceding presentation with endophthalmitis. These patients underwent 53 injections without PI (including bilaterally in Case 3) before developing endophthalmitis in the affected eye, which represents a 9.4% rate of endophthalmitis and a mean of 10.6 injections before endophthalmitis. When excluding injections without PI to the eye that did not go on to develop endophthalmitis (the left eye of Case 3), there were 44 injections without PI prior to development of endophthalmitis, which corresponds to an 11% rate of endophthalmitis and a mean of 8.8 injections without PI before endophthalmitis. At MEEI, the median interval between injection and presentation with endophthalmitis was 4 days and 3.5 days for those receiving and not receiving PI, respectively. At DMEI the median number of days between injection and presentation was 5 days in both groups. Patients presented for evaluation between 0 and 2 days after symptom onset.
All 5 patients who did not receive PI at the time of their injection were managed with tap-and-inject procedures and 3 received 5% PI antisepsis as part of the tap-and-inject procedure protocol, without complications. One patient (Case 5) did not receive PI at the time of tap and inject and required pars plana vitrectomy with intravitreal antibiotics within 24 hours because of worsening examination findings. This surgery was done without preoperative PI preparation. Cultures were positive in 4 of 5 cases: all grew coagulase-negative Staphylococcus . After resolution of endophthalmitis, 4 of the 5 patients went on to receive subsequent intravitreal injections, all with PI antisepsis and without allergic reactions. One of these 4 patients received a modified PI preparation, where a sterile cotton-tipped applicator soaked in PI was applied directly over the conjunctiva at the planned injection site immediately prior to injection, and ultimately went on to have another episode of endophthalmitis. The 1 patient who did not receive any post-endophthalmitis intravitreal injections did not have a reaction to PI use at the time of her tap and inject. The patients’ individual case histories are described below and summarized in Table 1 . Table 2 provides a summary of their injection histories.
| Patient | Age/Sex | Procedure | Reported Allergy | Alternative Agent Used | Days to Presentation | Management | Bacteriology | VA Before Endophthalmitis | VA on Presentation | VA at Last Follow-up |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 86/F | Intravitreal aflibercept | Iodine contrast—hives | Gatifloxacin 0.5% | 5 | T+I: Vancomycin 1 mg/0.1 mL and ceftazidime 2 mg/0.1 mL | S lugdunensis | 20/32 | HM | 20/80 |
| 2 | 71/F | Intravitreal ranibizumab | Iodine contrast—hives and respiratory distress | Gatifloxacin 0.5% | 33 | T+I: Vancomycin 1 mg/0.1 mL and ceftazidime 2 mg/0.1 mL | S epidermidis | 20/25+2 | 20/200 | 20/25 |
| 3 | 60/F | Intravitreal aflibercept | Topical iodine—eyelid swelling | Gatifloxacin 0.5% | 3 | T+I: Vancomycin 1 mg/0.1 mL and ceftazidime 2 mg/0.1 mL | – | 20/50 | HM | 20/125 |
| 4 | 65/F | Intravitreal bevacizumab | Topical iodine—eyelid swelling and hives | Shurclens | 4 | T+I: Vancomycin 1 mg/0.1 mL and ceftazidime 2 mg/0.1 mL | S epidermidis | 20/60 | HM | 20/70 |
| 5 | 86/F | Intravitreal bevacizumab | Anaphylaxis to iodine | Moxifloxacin 0.5% | 3 | T+I: Vancomycin 1 mg/0.1 mL and ceftazidime 2 mg/0.1 ml→PPV 48 hours later | S epidermidis | Unknown | HM | 20/160 |
| Patient | Procedural Eye | Injections Without Iodine | Injections With Iodine a | Total Injections a | Endophthalmitis Rate Without Iodine (%) |
|---|---|---|---|---|---|
| 1 | OD | 1 | 11 | 12 | 100 |
| 2 | OS | 12 | 6 | 18 | 8.33 |
| 3 | OU b | 36 | 72 c | 108 | 2.8 |
| 4 | OD | 1 | 2 | 3 | 100 |
| 5 | OS | 3 | 3 | 6 | 33 |
a Includes injections done after endophthalmitis.
b Patient developed endophthalmitis in the right eye.
Individual Case Histories
Case 1
An 86-year-old woman with a history of neovascular age-related macular degeneration (AMD) of the right eye, treated previously with a total of 12 intravitreal aflibercept injections, presented 5 days following the last injection with worsening eye pain and nausea. She was found to have hand motion (HM) visual acuity (decreased from 20/32 at last visit), a 1.5 mm hypopyon, and diffuse mobile vitreous debris. Unlike her prior 11 intravitreal injections, the patient’s most recent intravitreal injection was performed with topical preinjection gatifloxacin 0.5% instead of PI antisepsis owing to the patient’s report of a recent unknown adverse reaction to iodinated contrast material. She underwent an anterior chamber paracentesis (and vitreous aspirate, which was dry) and intravitreal injection of vancomycin (1 mg/0.1 mL) and ceftazidime (2 mg/0.1 mL) with use of 5% PI prior to the procedure without adverse reaction. Anterior chamber culture was positive for S lugdunensis , a coagulase-negative Staphylococcus (sensitive to vancomycin). Visual acuity improved to 20/80 at last follow-up. She did not receive further intravitreal injections.
Case 2
A 71-year-old woman with a history of neovascular AMD of the left eye, treated previously with photodynamic therapy and a total of 12 ranibizumab injections, presented 33 days after her last injection with 2 days of pain and photosensitivity. Visual acuity was 20/200, reduced from 20/25+ prior to injection. The eye had 4+ anterior chamber cells with a 0.75 mm hypopyon and vitreous opacities. She had received gatifloxacin 0.5% without PI prior to all previous injections owing to a history of respiratory distress and hives with intravenous iodine contrast. She underwent an intravitreal tap and inject with vancomycin (1 mg/0.1 mL) and ceftazidime (2 mg/0.1 mL); 5% PI was used prior to the procedure without adverse reaction. Vitreous cultures were positive for S epidermidis (sensitive to vancomycin). Visual acuity improved to 20/25 at last follow-up and she has received PI with her subsequent 6 injections without incident.
Case 3
A 62-year-old woman with a history of bilateral neovascular AMD presented 3 days after bilateral intravitreal aflibercept injections with pain, redness, and photophobia in the right eye. Vision was HM, a decline from 20/50 prior to injection. The right eye had 4+ anterior chamber cells, <1 mm hypopyon, and 3–4+ vitreous cell/debris. The patient underwent anterior chamber paracentesis (and vitreous tap, which was dry) and intravitreal injection of vancomycin (1 mg/0.1 mL) and ceftazidime (2 mg/0.1 mL) without PI owing to patient refusal. Anterior chamber culture was negative, without organisms on Gram stain. Visual acuity at last follow-up was 20/125. Prior to presentation she had received 31 intravitreal ranibizumab injections with PI in the right eye, after which she developed intolerable eyelid swelling and then refused PI. The subsequent treatment course included 20 intravitreal ranibizumab and 7 intravitreal aflibercept injections in the right eye (as well as 4 intravitreal ranibizumab and 5 aflibercept injections in the left eye) with gatifloxacin as a substitute for PI prep. After her episode of endophthalmitis, the patient was counseled about the importance of PI prep and was asked to reconsider injections with PI. She ultimately consented to a modified PI prep by which a sterile cotton-tipped applicator soaked in PI was applied to the conjunctiva directly over the planned injection site for 20–30 seconds. The injection protocol otherwise included topical lidocaine, no lid scrubs with iodine, and no masks. She had 20 injections of intravitreal aflibercept in the right eye (and 21 in the left eye) with this modified PI prep and developed another episode of endophthalmitis in the right eye. She was treated with tap and inject but insisted that the doctor use the modified PI prep. Vitreous cultures were positive for S epidermidis (sensitive to vancomycin).
Case 4
A 65-year-old woman with a history of neovascular AMD in the right eye presented 4 days following her first injection of intravitreal bevacizumab with pain, redness, and decreased vision from 20/60 to HM. Examination was notable for corneal edema, dispersed anterior chamber fibrin with 2 mm hypopyon, and moderate vitritis. Owing to reported allergy to topical PI consisting of swelling and hives, she had received Surclens (20% solution of poloxamer 188 in sterile water) as a substitute to PI antisepsis prior to injection. She underwent a tap and inject with vancomycin (1 mg/0.1 mL) and ceftazidime (2 mg/0.1 mL); 5% PI was used for this procedure. Vitreous cultures were positive for S epidermidis (sensitive to vancomycin). Visual acuity improved to 20/70 at last follow-up and she has received PI with her subsequent 2 injections without incident.
Case 5
An 86-year-old woman with a history of neovascular AMD in the right eye presented with floaters and decreased vision 3 days following her third intravitreal bevacizumab injection. She had HM vision, a hypopyon, and vitreous debris. Owing to patient-reported anaphylaxis to iodine-containing agents, she had been receiving topical moxifloxacin as a substitute to PI antisepsis prior to intravitreal injections. She underwent tap and inject with vancomycin (1 mg/0.1 mL) and ceftazidime (2 mg/0.1 mL); 5% PI was not used prior to the procedure because of the severity of the reported allergy and patient request. Vitreous cultures were positive for S epidermidis (sensitive to vancomycin). Owing to worsening of symptoms and visual acuity over the next 48 hours, she underwent pars plana vitrectomy with intravitreal injection of vancomycin, ceftazidime, and dexamethasone. Again, PI was not used as part of presurgical preparation. Visual acuity improved to 20/120 at last follow-up; she received 3 subsequent intravitreal injections of ranibizumab with PI preparation without complication.
Results
Injection protocols at both MEEI and DMEI were reviewed. The procedures were office-based at both institutions. To achieve anesthesia, proparacaine and lidocaine were employed. Topical lidocaine was the norm at DMEI, whereas physicians were divided evenly between topical and subconjunctival lidocaine use at MEEI. Topical PI was applied to the ocular surface; eyelid scrubs were not typically used. Masks and eyelid drapes were not used but speculum, gloves, and caliper use were the norm at both institutions. Topical antibiotics were typically not given after injections; however, 2 patients (Cases 2 and 3) who did not receive PI at the time of their injection received postinjection topical antibiotics, and 1 patient who did receive PI and went on to develop endophthalmitis (at MEEI) also received topical antibiotics. None of the patients with endophthalmitis (with or without PI use at the time of injection) at DMEI received postinjection antibiotics.
There were 30 046 injections performed at MEEI during the study period and 6 of these resulted in endophthalmitis, giving a 0.020% rate of postinjection endophthalmitis. There was a 0.018% rate of postinjection endophthalmitis for patients injected at DMEI (33 699 injections, 6 infections). The combined rate of endophthalmitis was 0.019% at both institutions. There were 21 postinjection endophthalmitis cases from outside doctors who referred the patients to MEEI and 9 cases referred to DMEI; however, because the total number of injections in the patient catchment area is unknown, the regional rate of endophthalmitis outside these institutions could not be calculated. In total, 42 cases of presumed intravitreal injection–related endophthalmitis were identified: 27 cases at MEEI and 15 at DMEI.
Five of the 42 patients (11.9%) were identified as having a history of self-reported prior allergic reaction to iodine-related agents and did not receive PI antisepsis at the time of the injection immediately preceding presentation with endophthalmitis. These patients underwent 53 injections without PI (including bilaterally in Case 3) before developing endophthalmitis in the affected eye, which represents a 9.4% rate of endophthalmitis and a mean of 10.6 injections before endophthalmitis. When excluding injections without PI to the eye that did not go on to develop endophthalmitis (the left eye of Case 3), there were 44 injections without PI prior to development of endophthalmitis, which corresponds to an 11% rate of endophthalmitis and a mean of 8.8 injections without PI before endophthalmitis. At MEEI, the median interval between injection and presentation with endophthalmitis was 4 days and 3.5 days for those receiving and not receiving PI, respectively. At DMEI the median number of days between injection and presentation was 5 days in both groups. Patients presented for evaluation between 0 and 2 days after symptom onset.
All 5 patients who did not receive PI at the time of their injection were managed with tap-and-inject procedures and 3 received 5% PI antisepsis as part of the tap-and-inject procedure protocol, without complications. One patient (Case 5) did not receive PI at the time of tap and inject and required pars plana vitrectomy with intravitreal antibiotics within 24 hours because of worsening examination findings. This surgery was done without preoperative PI preparation. Cultures were positive in 4 of 5 cases: all grew coagulase-negative Staphylococcus . After resolution of endophthalmitis, 4 of the 5 patients went on to receive subsequent intravitreal injections, all with PI antisepsis and without allergic reactions. One of these 4 patients received a modified PI preparation, where a sterile cotton-tipped applicator soaked in PI was applied directly over the conjunctiva at the planned injection site immediately prior to injection, and ultimately went on to have another episode of endophthalmitis. The 1 patient who did not receive any post-endophthalmitis intravitreal injections did not have a reaction to PI use at the time of her tap and inject. The patients’ individual case histories are described below and summarized in Table 1 . Table 2 provides a summary of their injection histories.
