The Role of Topical Antibiotic Prophylaxis to Prevent Endophthalmitis after Intravitreal Injection

The Role of Topical Antibiotic Prophylaxis to Prevent Endophthalmitis after Intravitreal Injection Philip Storey, MD, MPH, 1 Michael Dollin, MD, 1 John Pitcher, MD, 1 Sahitya Reddy, BA, 2 Joseph Vojtko, BA, 2 James Vander, MD, 1 Jason Hsu, MD, 1 Sunir J. Garg, MD, 1 for the Post-Injection Endophthalmitis Study Team* Objective: To compare the incidence of endophthalmitis after intravitreal injection with and without topical postinjection antibiotic prophylaxis. Design: Retrospective case-control study. Participants: All patients treated with intravitreal injection of ranibizumab, bevacizumab, or aflibercept for a variety of retinal vascular diseases at a single, large retina practice between January 1, 2009, and October 1, 2012, were included. Methods: The total numbers of patients and injections were determined from a review of billing code and practice management records. Endophthalmitis cases were determined from billing records and from an infection log. All cases of endophthalmitis were confirmed with chart review. A 28-month period when topical antibiotics were prescribed after intravitreal injection was compared with a 9-month period when topical antibiotics were not prescribed. Patients treated during an 8-month transition period were excluded to allow for the conversion of antibiotic prescription practices. Main Outcome Measures: Incidence of endophthalmitis, visual acuity outcomes, and microbial spectrum. Results: During the study period, a total of 117 171 intravitreal injections were performed (57 654 injections during the topical antibiotic period, 24 617 during the transition period, and 34 900 during the no-antibiotic period), with a total of 44 cases of suspected endophthalmitis (0.038%; 1 in 2663 injections), 17 of which showed culture-positive results (0.015%; 1 in 6892 injections). During the 28-month topical antibiotic period, there were 28 cases of suspected endophthalmitis (0.049%; 1 in 2059 injections), 10 of which showed culturepositive results (0.017%; 1 in 5765 injections). During the 9-month no-antibiotic period, there were 11 cases of suspected endophthalmitis (0.032%; 1 in 3173 injections), 4 of which showed culture-positive results (0.011%; 1 in 8725 injections). Topical antibiotic use was associated with a trend toward increased risk of suspected endophthalmitis (odds ratio [OR], 1.54; 95% confidence interval [CI], 0.77e3.10) and culture-positive endophthalmitis (OR, 1.51; 95% CI, 0.47e4.83). Conclusions: The incidence of endophthalmitis after intravitreal injection is low. Using postinjection topical antibiotic drops does not reduce the risk of endophthalmitis developing and is associated with a trend toward higher incidence of endophthalmitis. Ophthalmology 2014;121:283-289 ª 2014 by the American Academy of Ophthalmology. *Group members listed online in Appendix 1 (available at http://aaojournal.org). The use of intravitreal injections for eye disease has increased dramatically because of the efficacy of several medications, including steroids and antievascular endothelial growth factor (VEGF) agents. Although uncommon, postinjection endophthalmitis can cause significant ocular morbidity. 1 Prophylactic measures, including the use of various topical antibiotics, have been used to reduce the incidence of endophthalmitis. Topical antibiotics long have been the standard of care after ocular surgery, and this practice was carried over to ocular injections. Although topical antibiotics reduce conjunctival bacterial growth 2 and may have a possible synergistic effect when combined with povidoneeiodine, 3 no randomized trials have demonstrated a reduction in postprocedure or postinjection endophthalmitis through use of prophylactic antibiotic drops. 4e6 Recent evidence suggests that topical antibiotics may even be harmful by increasing antibiotic-resistant bacterial strains 7,8 and may increase the risk of endophthalmitis. 9 The previous studies were relatively small and may not have been powered adequately to detect a difference in endophthalmitis incidence. Many physicians continue to prescribe antibiotics after intravitreal injection. 10 This study evaluates the role of topical antibiotic prophylaxis after intravitreal injection of anti-vegf agents as well as the microbial spectrum after injection in the largest series of intravitreal injections to date. Ó 2014 by the American Academy of Ophthalmology Published by Elsevier Inc. ISSN 0161-6420/14/$ - see front matter http://dx.doi.org/10.1016/j.ophtha.2013.08.037 283

Ophthalmology Volume 121, Number 1, January 2014 Methods Overview This single-center retrospective, comparative, case-control study was approved by the Wills Eye Institute Institutional Review Board. As part of an ongoing infection surveillance program, the authors prospectively recorded endophthalmitis cases occurring after intravitreal injection secondary to intravitreal injection of bevacizumab (Genentech, South San Francisco, CA), ranibizumab (Genentech), triamcinolone, or aflibercept (Regeneron, Tarrytown, NY). The endophthalmitis log and billing records were used to identify retrospectively all cases of endophthalmitis treated with vitreous tap and injection between January 1, 2009, and October 1, 2012, at a single retina practice. The total number of intravitreal injections performed during the period was determined from billing data. Charts of all patients who were treated for endophthalmitis were reviewed. Recorded data included date of causative injection; date of tap and injection; visual acuity before causative injection, at time of tap and injection, at 3 months after the procedure, and at 6 months after the procedure; type of injection; lot number; underlying diagnosis; culture results; and antibiotic resistance. Inclusion and Exclusion Criteria All eyes with presumed infectious endophthalmitis following intravitreal injection of an anti-vegf medication were included. Endophthalmitis was defined as any case in which clinical suspicion was high enough to warrant tap and inject (typically decreased visual acuity, pain, 2þ anterior chamber cell, and vitreitis). Eyes with presumed inflammatory endophthalmitis treated primarily with topical steroids without tap-and-inject treatment were excluded from analysis. Injection Technique All injections were performed in office-based settings. Eyes were prepped in a standardized method with a topical nonviscous anesthetic, one topical antibiotic drop during the initial 28 months of the study period, topical 5% povidoneeiodine (Betadine 5%; Alcon Labs, Fort Worth, TX), followed by another drop of topical anesthetic and another drop of 5% povidoneeiodine before injection. Rarely, a subconjunctival 2% lidocaine injection was administered before the second administration of topical anesthetic and 5% povidoneeiodine. A sterile drape and eyelash preparation were not used. Injection with a 30- or 31-gauge needle was performed 3.5 to 4.0 mm from the limbus. Physicians individually determined use of a bladed lid speculum, conjunctival displacement before injection, and superior versus inferior injection site. Endophthalmitis Treatment Protocol All eyes with presumed infectious endophthalmitis immediately underwent a pars plana vitreous tap using a 25- or 27-gauge needle with attempted aspiration and subsequent injection of intravitreal antibiotics. If the physician was unable to obtain vitreous fluid, an aqueous tap was performed. Patients received intravitreal vancomycin (1 mg/0.1 ml) and ceftazidime (2 mg/0.1 ml). Intravitreal amikacin (400 mg/0.1 ml) was substituted for ceftazidime for patients with penicillin allergy. Subsequent intravitreal antibiotics were modified based on culture sensitivities. Patients variably were prescribed atropine sulfate 1% drops twice daily as well as fortified vancomycin (25 mg/ml), fortified tobramycin (15 mg/ml), and prednisolone acetate 1% drops every hour. Patients were followed up daily until evidence of clinical improvement was apparent. At that point, the drops were tapered slowly and the follow-up interval was extended. Transition Period Before May 2011, all patients were prescribed postinjection topical antibiotics to use 4 times daily for 4 days. As early as June 2011, some physicians stopped using antibiotic prophylaxis, and on September 1, 2011, retinal physicians practicing at Wills Eye Institute adopted a practice-wide policy stopping use of postinjection topical antibiotics. We considered May 1, 2011, through December 31, 2011, to be a transition period during which physicians and patients changed prophylaxis strategies. A review of all endophthalmitis cases confirmed that all patients from January 1, 2009, through April 30, 2011, received postinjection antibiotics, whereas all patients from January 1, 2012, through October 1, 2012, did not receive postinjection antibiotics. We did not include injections or endophthalmitis cases during the transition period in our analysis of endophthalmitis incidence. We did, however, include transition period cases in our analysis of visual outcomes and bacterial spectrum. Outcomes The primary outcome was occurrence of endophthalmitis after intravitreal injection. Secondary outcomes were microbial spectrum and clinical outcomes including return to baseline visual acuity (plus or minus 2 lines of Snellen acuity), final visual acuity of counting fingers or worse, and the presence of pain, vitreitis, or hypopyon on initial presentation. Endophthalmitis results were considered culture positive if there were positive gram stain or positive growth results on culture plates. Clinical variables were analyzed using Excel (Microsoft, Redmond, WA), and statistical analysis was performed using GraphPad Software (GraphPad, La Jolla, CA). Results Effect of Antibiotics Between January 1, 2009, and October 1, 2012, a total of 117 171 intravitreal injections (71 791 ranibizumab, 44 007 bevacizumab, and 1373 aflibercept) were performed, and a total of 44 patients with suspected endophthalmitis after intravitreal injection underwent vitreous tap with antibiotic injection (0.038%; 1 in 2663 injections). Seventeen cases showed culture-positive results (0.015%; 1 in 6892 injections). Overall rates of suspected endophthalmitis were 24 in 71 791 injections for ranibizumab (0.033%; 1 in 2991 injections); 20 in 44 007 injections for bevacizumab (0.045%; 1 in 2200 injections); and 0 in 1373 injections for aflibercept. Overall rates of culture-positive endophthalmitis were 12 in 71 791 for ranibizumab (0.017%; 1 in 5983 injections), 5 in 44 007 for bevacizumab (0.011%; 1 in 8801 injections), and 0 in 1373 for aflibercept. There was no statistically significant difference in suspected or culture-positive endophthalmitis among the various agents. During the 28-month period when postinjection topical antibiotics were prescribed, 57 654 injections (36 781 ranibizumab, 20 873 bevacizumab, and 0 aflibercept) were administered and 28 cases of suspected endophthalmitis occurred (0.049%; 1 in 2059 injections), 10 of which showed culture-positive results (0.017%; 1 in 5765 injections; Table 1). Twenty-four eyes received anti-vegf injection for neovascular age-related macular degeneration (AMD) and 4 received injection for macular edema secondary to retina vein occlusion. Causative organisms included 4 cases of Streptococcus viridans, 2 cases of Enterococcus, 2 cases of coagulasenegative Staphylococcus, and 1 case each of Staphylococcus aureus and Lactobacillus. 284

Storey et al Antibiotics to Prevent Endophthalmitis Table 1. Suspected Endophthalmitis Cases and Antibiotic Prophylaxis Strategies Prophylactic Topical Antibiotics No Prophylactic Antibiotics Intravitreal Medication Injections Cases (Incidence) Injections Cases (Incidence) Odds Ratio (95% Confidence Interval) Ranibizumab 36 781 16 (0.044%) 20 832 7 (0.034%) 1.29 (0.53e3.15) Bevacizumab 20 873 12 (0.057%) 12 697 4 (0.032%) 1.83 (0.59e5.66) Aflibercept 0 0 1371 0 d Total 57 654 28 (0.049%) 34 900 11 (0.032%) 1.54 (0.77e3.10) During the 8-month transition period, 24 617 injections (14 178 ranibizumab, 10 437 bevacizumab, and 2 aflibercept) were performed. Five cases of suspected endophthalmitis occurred (0.020%; 1 in 4923), and 3 showed culture-positive results (0.012%; 1 in 8206 injections). The 3 cases in which topical antibiotics were used all grew coagulase-negative Staphylococcus. No growth was found in the 2 cases in which no topical antibiotics were prescribed. Three patients were being treated for neovascular AMD, whereas 2 patients received treatment for diabetic macular edema. During the 9-month period in which postinjection antibiotics were not prescribed, 34 900 injections (20 832 ranibizumab, 12 697 bevacizumab, and 1371 aflibercept) were administered and 11 cases of suspected endophthalmitis occurred (0.032%; 1 in 3173 injections), 4 of which showed culture-positive results (0.011%; 1 in 8725 injections; Table 1). All 11 cases of suspected endophthalmitis received anti-vegf injection for neovascular AMD. Causative organisms included 2 cases of S. viridans, 1 case of S. aureus, and 1 case of nonspeciated gram-positive cocci on gram stain. Compared with the period in which no antibiotics were used, the use of postinjection topical antibiotics was associated with a trend toward increased incidence of both suspected endophthalmitis (odds ratio, 1.54; 95% confidence interval, 0.77e3.10; Table 1) and culture-positive endophthalmitis (odds ratio, 1.51; 95% confidence interval, 0.47e4.83; Table 2). Clinical Outcomes Overall, patients with presumed endophthalmitis reported pain, redness, or decreased vision an average of 3.7 days after injection (range, 1e11 days). Patients prescribed topical antibiotics sought treatment an average of 3.7 days after injection versus an average of 3.5 days for patients not prescribed antibiotics (P ¼ 0.82). Regardless of antibiotic prophylaxis strategy, culture-positive cases sought treatment an average of 3.6 days after injection, compared with 3.7 days for culture-negative cases (P ¼ 0.84). One culturenegative case was excluded from analysis because the patient s nursing home delayed seeking care until 17 days after injection. Postinjection antibiotic use did not alter the clinical presentation of patients with suspected or culture-positive endophthalmitis. For suspected endophthalmitis cases, all patients had decreased vision and most reported pain (29/31 postintravitreal injection antibiotic eyes vs. 13/13 no postintravitreal injection antibiotic eyes; P ¼ 1.0), vitreitis (23/31 postintravitreal injection antibiotic eyes vs. 7/13 no postintravitreal injection antibiotic eyes; P ¼ 0.29), and hypopyon (24/31 postintravitreal injection antibiotic eyes vs. 7/13 no postintravitreal injection antibiotic eyes; P ¼ 0.16). Clinical presentation of patients with culture-positive endophthalmitis did not differ with the use of topical antibiotics: all patients had decreased vision and most reported pain (11/13 postintravitreal injection antibiotic eyes vs. 4/4 no postintravitreal injection antibiotic eyes; P ¼ 1.0), vitreitis (12/13 postintravitreal injection antibiotic eyes vs. 3/4 no postintravitreal injection antibiotic eyes; P ¼ 0.43), and hypopyon (11/13 postintravitreal injection antibiotic eyes vs. 3/4 no postintravitreal injection antibiotic eyes; P ¼ 1.0). We found no statistically significant differences in clinical presentation between culture-positive versus culturenegative cases. Visual Outcomes Mean follow-up for all suspected endophthalmitis cases was 20.7 months. Two of 44 patients (1 culture positive, 1 culture negative) had follow-up of less than 1 month and were excluded from visual outcome analysis. Most eyes (54.8%; 23/42) returned to baseline visual acuity (2 lines of visual acuity) by 3 months after injection. Four additional cases returned to baseline visual acuity by 6 months. We found that use of postinjection topical antibiotics did not affect return to baseline visual acuity at 6 months (16/29 postintravitreal injection antibiotic eyes vs. 8/13 no postintravitreal injection antibiotic eyes; P ¼ 0.75). There was also no difference between the 2 groups in the proportion of patients with visual acuity of counting fingers or worse at 6 months (11/29 postintravitreal injection antibiotic eyes vs. 6/13 no postintravitreal injection antibiotic eyes up; P ¼ 0.74). Visual acuity outcomes were significantly worse for culturepositive cases compared with culture-negative cases, regardless of topical antibiotic use. At 6 months, only 18.8% of culturepositive cases (3/16) returned to baseline visual acuity compared with 80.8% (21/26) of culture-negative cases (P < 0.001). At 6 months, 62.5% of culture-positive cases (10/16) had vision of counting fingers or worse compared with 26.9% (7/26) of culturenegative cases (P ¼ 0.029). Visual outcomes of culture-positive and culture-negative cases are displayed in Tables 3 and 4, respectively. Table 2. Culture-Positive Endophthalmitis Cases and Antibiotic Prophylaxis Strategies Prophylactic Topical Antibiotics No Prophylactic Antibiotics Intravitreal Medication Injections Cases (Incidence) Injections Cases (Incidence) Odds Ratio (95% Confidence Interval) Ranibizumab 36 781 8 (0.022%) 20 832 3 (0.014%) 1.51 (0.40e5.69) Bevacizumab 20 873 2 (0.010%) 12 697 1 (0.0079%) 1.22 (0.11e13.4) Aflibercept 0 0 1371 0 d Total 57 654 10 (0.017%) 34 900 4 (0.011%) 1.51 (0.47e4.83) 285

Patient No. Agent Ophthalmology Volume 121, Number 1, January 2014 Postinjection Topical Antibiotics Used Table 3. Visual Acuity Outcomes for Culture-Positive Cases At Injection At Presentation Visual Acuity At 3 Months after Endophthalmitis At 6 Months after Endophthalmitis Culture Results 1 Ranibizumab Yes 20/300 HM NLP NLP Streptococcus viridans 2 Bevacizumab Yes 20/200 HM 20/200 CF Coagulase- negative 3 Ranibizumab Yes 20/100 NLP NLP NLP Enterococcus faecalis 4 Ranibizumab Yes 20/200 LP 20/400 20/400 Enterococcus faecalis 5 Ranibizumab Yes 20/200 HM CF 20/400 Staphylococcus aureus 6 Ranibizumab Yes 20/40 20/400 CF CF Streptococcus mitis 7 Ranibizumab Yes 20/40 CF CF HM Streptococcus veridans 8 Ranibizumab Yes 20/200 HM LP HM Streptococcus mitis 9 Bevacizumab Yes 20/100 HM Lost to follow-up Lost to follow-up Lactobacillus 10 Ranibizumab Yes 20/30 HM 20/80 20/60 Coagulase- negative 11 Ranibizumab Yes (transition period) 20/30 HM HM 20/50 Coagulase- negative 12 Bevacizumab Yes (transition period) 20/20 HM 20/70 20/40 Coagulase- negative 13 Bevacizumab Yes (transition period) 20/40 HM 20/50 HM Coagulase- negative 14 Bevacizumab No 20/80 HM NLP NLP Streptococcus salivarius 15 Ranibizumab No CF LP LP LP Streptococcus sanguis 16 Ranibizumab No 20/20 HM 20/70 20/50 Nondifferentiated gram-positive cocci 17 Ranibizumab No 20/40 HM CF CF Staphylococcus aureus CF ¼ counting fingers; HM ¼ hand movements; LP ¼ light perception; NLP ¼ no light perception. Discussion In this single-center, retrospective, case-control study, we compared endophthalmitis incidence during a 28-month period when topical antibiotics were prescribed with that during a 9-month period when no antibiotics were prescribed, separated by an 8-month transition period as providers and patients altered prophylaxis practices. We found that postinjection topical antibiotics did not decrease endophthalmitis incidence and that the clinical presentation and visual outcomes of patients with suspected endophthalmitis were similar regardless of whether postinjection topical antibiotics were used initially. Previous studies comparing topical antibiotics prescribed for several days after injection with no topical antibiotics have not found statistically significant differences in endophthalmitis rates, although all had relatively small sample sizes. Bhatt et al 4 concluded that there was no difference in rates of suspected endophthalmitis with or without topical antibiotics (5/2287 or 0.22% with antibiotics vs. 5/2480 or 0.20% no antibiotics; P ¼ 0.90). Cheung et al 9 concluded that the overall rate of intravitreal injection-related endophthalmitis is greater with the use of topical antibiotics compared with no antibiotics (5/8259 or 0.061% with 5 days of antibiotics vs. 2/5266 or 0.038% with no antibiotics; P ¼ 0.57). Cheung et al also concluded that if only culture-positive endophthalmitis cases were considered, the use of topical antibiotics showed lower endophthalmitis rates compared with patients receiving no antibiotics (1/8259 or 0.012% with 5 days of antibiotics vs. 2/5266 or 0.038% no antibiotics; P ¼ 0.32). However, the study s sample size was small, leading some investigators to state that the conclusions may be the result of a random sampling error. 11 Given the increasing number of studies suggesting that topical antibiotics after intravitreal injection do not prevent endophthalmitis, some investigators believe that there is no need for topical antibiotic use after intravitreal injection. 11,12 In 2012, the American Academy of Ophthalmology joined the Choosing Wisely campaign initiated by the American Board of Internal Medicine Foundation to create a list of Five Things Physicians and Patients Should Question. 13 The American Academy of Ophthalmology listed [not] routinely providing antibiotics before or after intravitreal injections as a possible intervention to eliminate unnecessary costs. 14 However, there are variable data on the usefulness of postinjection antibiotics to prevent endophthalmitis, and therefore no current guidelines recommending for or against their use. Our study aimed to evaluate the effect of prophylactic antibiotics in the largest series to date. Our study confirmed the finding of Cheung et al 9 of greater incidence of endophthalmitis with the use of topical antibioticsdalthough neither study found the increased risk to be statistically significant. Although our infection rates were lower, we found a similarly sized increased risk of endophthalmitis with the use of topical antibiotics (odds ratio, 1.54 in the current study vs. 1.59 in the study by Cheung et al). In contrast to Cheung et al, we found that topical antibiotics also were associated with a trend 286

Storey et al Antibiotics to Prevent Endophthalmitis Patient No. Agent Table 4. Visual Acuity Outcomes for Culture-Negative Cases Postinjection Topical Antibiotics Used At Injection At Presentation Visual Acuity At 3 Months after Endophthalmitis At 6 Months after Endophthalmitis 18 Bevacizumab Yes 20/100 CF 20/80 20/200 19 Bevacizumab Yes 20/70 HM CF 20/70 20 Ranibizumab Yes 20/50 HM 20/100 20/100 21 Bevacizumab Yes 20/70 HM 20/100 20/50 22 Bevacizumab Yes 20/200 HM CF CF 23 Bevacizumab Yes 20/200 HM 20/400 20/200 24 Bevacizumab Yes 20/200 HM 20/300 20/200 25 Bevacizumab Yes 20/30 20/250 20/30 20/30 26 Ranibizumab Yes 20/30 20/400 20/50 20/60 27 Bevacizumab Yes 20/30 HM 20/30 20/30 28 Bevacizumab Yes CF CF CF CF 29 Ranibizumab Yes 20/200 HM Lost to follow-up Lost to follow-up 30 Ranibizumab Yes 20/400 CF 20/400 20/400 31 Ranibizumab Yes 20/40 LP CF CF 32 Ranibizumab Yes 20/60 HM 20/100 20/100 33 Bevacizumab Yes 20/30 CF 20/30 20/40 34 Ranibizumab Yes 20/60 20/200 20/60 20/30 35 Ranibizumab Yes 20/125 HM LP NLP 36 Bevacizumab No (transition period) CF CF 20/60 20/60 37 Bevacizumab No (transition period) CF CF 20/400 20/400 38 Bevacizumab No 20/100 20/200 20/50 20/60 39 Ranibizumab No 20/40 CF 20/70 20/70 40 Ranibizumab No 20/400 CF CF CF 41 Bevacizumab No 20/40 CF 20/40 CF 42 Bevacizumab No 20/40 CF CF 20/60 43 Ranibizumab No 20/50 20/80 20/50 20/30 44 Ranibizumab No CF HM CF CF CF ¼ counting fingers; HM ¼ hand movements; LP ¼ light perception; NLP ¼ no light perception. toward increased risk of culture-positive endophthalmitis. We found no difference in rates of suspected or culturepositive endophthalmitis between anti-vegf agents. Several possibilities may explain the trend toward increased endophthalmitis rates with topical antibiotic use. Several studies have shown that topical antibiotics lead to an increase in resistant organisms. Milder et al 8 showed that repeated use of topical fluoroquinolone drops after intravitreal injection resulted in a significantly increased rate of bacterial resistance (63.6%) compared with control eyes (32.1%). Kim et al 15 demonstrated substantial levels of resistance to third- and fourth-generation fluoroquinolones as well as multidrug resistance in patients treated with topical antibiotics after multiple intravitreal injections. Cultures from eyes with endophthalmitis have shown a higher incidence of antibiotic resistance among bacterial isolates, 16,17 and resistant bacterial strains have been found to cause more severe ocular inflammation than sensitive strains. 18 Several recent studies have found that organisms typically associated with oral flora are more prevalent with endophthalmitis occurring after intravitreal injection than after other vitreoretinal procedures. 1,7,19 Our study also found that oropharyngeal organisms were more common after intravitreal injection than in reports of other eye procedures. 20 For the 17 culture-positive cases of endophthalmitis, 6 cases grew Streptococcus species (35.3%). There has been much speculation concerning the potential increased association of intravitreal injections with oral flora. Oropharyngeal droplet transmission may be the cause of this increased risk of Streptococcus infection after injection. One difference between intravitreal injections and other ocular procedures is that injections often are performed in the office, often with no masking of the physician, patient, or technician. In contrast, all persons in the operating room are masked and the patient is draped. A recent study found that rates of postintravitreal injection endophthalmitis were lower when the procedure was performed in the operating room as compared with in the office. 21 One suggested strategy to reduce droplet transmission is talking cessation. However, this has the inherent limitation of prohibiting communication between the physician and patient during the procedure and could increase the potential for incorrect site procedures. Our study was not designed to evaluate the role of other strategies such as wearing protective masks during intravitreal procedures. Strengths of this study include the large number of intravitreal injections performed by a single retina group over a fairly short period. A potential limitation of this study is its retrospective, case-control nature. However, identification of endophthalmitis cases did not depend on physician recall. All cases were identified using a combination of a prospective endophthalmitis log, as well as 287

Ophthalmology Volume 121, Number 1, January 2014 from billing records and chart review of all patients treated for endophthalmitis resulting from any cause during the study period. Adherence to antibiotic drop use after injection was not assessed during the period when antibiotics were prescribed. If antibiotics do increase rates of endophthalmitis, nonadherence to topical antibiotics would tend to underestimate the incidence of endophthalmitis in the antibiotic use group. Conversely, some patients could have continued to use antibiotic drops even after being instructed not to do so. An 8-month transition period was included in the study design to limit this potential source of bias. Additionally, a patient could have developed endophthalmitis and sought treatment elsewhere, but given the tertiary care nature of our practice, this is less likely, and we know of no instances in which this occurred. Finally, retina specialists involved in the study knew that topical antibiotic prophylaxis was discontinued at a specific time. It is possible that physicians increased precautions (e.g., extra povidoneeiodine, refraining from talking, change in speculum use) after topical antibiotic use was discontinued. However, no retina specialists reported increasing their customary precautions after stopping prophylactic antibiotic use. Furthermore, physicians were unaware that a study would evaluate the effect of prophylactic antibiotics, making an unmasking bias less likely. Although the probability of developing endophthalmitis after intravitreal injection is low, treatment of AMD, retinal vein occlusions, and diabetic macular edema can require years of repeated injections. Consequently, ophthalmologists have attempted to prevent infection through a number of measures, including prophylactic topical antibiotics. Our study adds to the current body of literature indicating that postinjection topical antibiotics do not decrease rates of endophthalmitis. Given evidence suggesting potential harm of topical antibiotics and a potential increase in endophthalmitis rates with their use, we recommend against the routine use of topical antibiotics after intravitreal injection. References 1. Shah CP, Garg SJ, Vander JF, et al. Post-Injection Endophthalmitis (PIE) Study Team. Outcomes and risk factors associated with endophthalmitis after intravitreal injection of anti-vascular endothelial growth factor agents. Ophthalmology 2011;118:2028 34. 2. Moss JM, Sanislo SR, Ta CN. A prospective randomized evaluation of topical gatifloxacin on conjunctival flora in patients undergoing intravitreal injections. Ophthalmology 2009;116:1498 501. 3. Ta CN, Egbert PR, Singh K, et al. Prospective randomized comparison of 3-day versus 1-hour preoperative ofloxacin prophylaxis for cataract surgery. Ophthalmology 2002;109: 2036 40. discussion 2040e1. 4. Bhatt SS, Stepien KE, Joshi K. Prophylactic antibiotic use after intravitreal injection: effect on endophthalmitis rate. Retina 2011;31:2032 6. 5. Bhavsar AR, Ip MS, Glassman AR, et al. DRCRnet and the SCORE Study Groups. The risk of endophthalmitis following intravitreal triamcinolone injection in the DRCRnet and SCORE clinical trials. Am J Ophthalmol 2007;144:454 6. 6. Bhavsar AR, Googe JM Jr, Stockdale CR, et al. Diabetic Retinopathy Clinical Research Network. Risk of endophthalmitis after intravitreal drug injection when topical antibiotics are not required: the Diabetic Retinopathy Clinical Research Network laser-ranibizumabtriamcinolone clinical trials. Arch Ophthalmol 2009;127: 1581 3. 7. Kim SJ, Toma HS. Antimicrobial resistance and ophthalmic antibiotics: 1-year results of a longitudinal controlled study of patients undergoing intravitreal injections. Arch Ophthalmol 2011;129:1180 8. 8. Milder E, Vander J, Shah C, Garg S. Changes in antibiotic resistance patterns of conjunctival flora due to repeated use of topical antibiotics after intravitreal injection. Ophthalmology 2012;119:1420 4. 9. Cheung CS, Wong AW, Lui A, et al. Incidence of endophthalmitis and use of antibiotic prophylaxis after intravitreal injections. Ophthalmology 2012;119:1609 14. 10. Green-Simms AE, Ekdawi NS, Bakri SJ. Survey of intravitreal injection techniques among retinal specialists in the United States. Am J Ophthalmol 2011;151:329 32. 11. Kulkarni AA, Altaweel MM, Tewari A. Comment on: Cheung CS, Wong AW, Lui A, et al. Incidence of endophthalmitis and use of antibiotic prophylaxis after intravitreal injections. Ophthalmology 2012;119:1609e14. In: American Society of Retina Specialists October Literature Roundup. April 2012. Available at: http://www.asrs.org/patients/retina-news/literatureroundup/91/october-literature-roundup. Accessed August 18, 2013. 12. Chen RW, Rachitskaya A, Scott IU, Flynn HW. Is the use of topical antibiotics for intravitreal injections the standard of care or are we better off without antibiotics? JAMA Ophthalmol 2013;131:840 2. 13. Cassel CK, Guest JA. Choosing wisely: helping physicians and patients make smart decisions about their care. JAMA 2012;307:1801 2. 14. Parke DW II, Coleman AL, Rich WL III, Lum F. Choosing wisely: five ideas that physicians and patients can discuss. Ophthalmology 2013;120:443 4. 15. Kim SJ, Toma HS, Midha NK, et al. Antibiotic Resistance of Conjunctiva and Nasopharynx Evaluation Study: a prospective study of patients undergoing intravitreal injections. Ophthalmology 2010;117:2372 8. 16. Miller D, Flynn PM, Scott IU, et al. In vitro fluoroquinolone resistance in staphylococcal endophthalmitis isolates. Arch Ophthalmol 2006;124:479 83. 17. Yin VT, Weisbrod DJ, Eng KT, et al. Antibiotic resistance of ocular surface flora with repeated use of a topical antibiotic after intravitreal injection. JAMA Ophthalmol 2013;131: 456 61. 18. Hooper DC. Fluoroquinolone resistance among gram-positive cocci. Lancet Infect Dis 2002;2:530 8. 19. McCannel CA. Meta-analysis of endophthalmitis after intravitreal injection of anti-vascular endothelial growth factor agents: causative organisms and possible prevention strategies. Retina 2011;31:654 61. 20. Chen E, Lin MY, Cox J, Brown DM. Endophthalmitis after intravitreal injection: the importance of viridans streptococci. Retina 2011;31:1525 33. 21. Abell RG, Kerr NM, Allen P, Vote BJ. Intravitreal injections: is there benefit for a theatre setting? Br J Ophthalmol 2012;96: 1474 8. 288

Storey et al Antibiotics to Prevent Endophthalmitis Footnotes and Financial Disclosures Originally received: May 29, 2013. Final revision: August 24, 2013. Accepted: August 27, 2013. Available online: October 21, 2013. Manuscript no. 2013-861. 1 The Retina Service of Wills Eye Hospital, Mid Atlantic Retina, The Retina Service of Wills Eye Hospital, Thomas Jefferson University, Philadelphia, Pennsylvania. 2 Department of Ophthalmology, Thomas Jefferson University, Philadelphia, Pennsylvania. *A full listing of The Post-Injection Endophthalmitis (PIE) Study Team is available online in appendix 1 (http://aaojournal.org). All members of the PIE Study Team are from The Retina Service of Wills Eye Hospital, MidAtlantic Retina, Thomas Jefferson University, Philadelphia, PA. Presented at: Wills Eye Hospital Annual Conference, 2013; Retina Society Annual Meeting, 2013; and as a poster at: Association for Research in Vision and Ophthalmology Annual Meeting, 2013; and American Academy of Ophthalmology Annual Meeting, November 2013, New Orleans, LA. Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article. Correspondence: Sunir J. Garg, MD, Mid Atlantic Retina, The Retina Service of Wills Eye Hospital, Thomas Jefferson University, 840 Walnut Street, Suite 1020, Philadelphia, PA 19107. E-mail: sunirgarg@yahoo.com. 289