Pathogens and antibiotic sensitivities in endophthalmitis

Abstract

BACKGROUND

Antibiotic prophylaxis in cataract surgery is intended to minimize endophthalmitis. We describe pathogenic organisms, antibiotic sensitivities, and antibiotic prophylaxis in culture-proven endophthalmitis cases.

DESIGN

Retrospective consecutive case series, community-based setting.

SAMPLES

215 cases of endophthalmitis after cataract surgery performed during 2007–2012 in Kaiser Permanente, California.

METHODS AND MAIN OUTCOME MEASURES

Descriptive analysis of isolated organisms and antibiotic sensitivities in relation to antibiotic prophylaxis in culture-proven endophthalmitis cases.

RESULTS

The majority of culture-confirmed organisms (N=83) were Gram-positive (96%), most notably coagulase-negative Staphylococci (CoNS) (N=34, 52%), of which all that underwent testing were sensitive to vancomycin (N=32). Among 19 cases that had received only topical antibiotic prophylaxis, 7 (37%) were resistant to the antibiotic given: 50% of cases (5 of 10 isolates) that had received ofloxacin were resistant to this antibiotic, 40% (2 of 5 isolates) that had received gatifloxacin were resistant. In contrast, 100% of cases (N=4) that had received aminoglycosides were susceptible. Few culture-confirmed cases occurred in patients who received intracameral antibiotic (N=4).

CONCLUSIONS

In cases where fluoroquinolones were administered as antibiotic prophylaxis, isolates demonstrated a degree of bacterial resistance. The majority of endophthalmitis cases isolated following topical antibiotic prophylaxis only and were attributed to Gram-positive organisms, while few occurred in association with intracameral antibiotic.

INTRODUCTION

Historically, the incidence of endophthalmitis after cataract surgery has ranged from 0.03% to 0.70%.^1,2 The most common causative organisms are Gram-positive, such as coagulase-negative Staphylococcus (CoNS), Streptococcus viridans or Staphylococcus aureus.³ Gram-negative organisms,⁴ such as Pseudomonas or Haemophilus, are less common; fungi are rare.^1,4,5

The injection of intracameral antibiotics has been shown to decrease the rate of endophthalmitis in cataract surgery.^1,6–12 Topical antibiotics are intended to reduce conjunctival bacterial load which is predominantly composed of CoNS.^4,13,14 We showed a 42% reduction in endophthalmitis associated with intracameral antibiotic compared with topical antibiotic, whereas no use of antibiotic prophylaxis was associated with a 95% increase in risk compared with use of topical antibiotic.¹² The use of topical antibiotics alone, however, is associated with concerns about inadequate bacterial eradication,^13,15,16 inadequate minimum inhibitory concentration,^17,18 and insufficient effect on lowering the risk of endophthalmitis.^6,19,20 Nevertheless, this method of antibiotic administration is still more common in the U.S.A.²¹

We described the pathogenic organisms, their antibiotic sensitivities, and type of antibiotic prophylaxis in 215 endophthalmitis cases diagnosed in Kaiser Permanente California from 2007 to 2013 to measure the effectiveness of antibiotic agent and route of administration.

METHODS

The study was approved by the Kaiser Foundation Research Institute Institutional Review Board. Detailed methods are provided elsewhere ¹² and summarized here.

Setting

Kaiser Permanente is a closed, staff model, integrated healthcare delivery system with capitated payment that provides comprehensive care to 6.5 million Californians, representing one-third of covered individuals in the catchment area. The health plan uses an electronic health record to store detailed clinical information. Clear cornea phacoemulsification using the Alcon Infiniti is performed at 21 surgical centers in Northern California and at 17 surgical centers in Southern California. Surgeons have autonomy in choosing their chemoprophylaxis regimen for infection prevention.

Study Population

This study is a retrospective consecutive case series of endophthalmitis cases that occurred within 315,246 eligible non-complex phacoemulsification procedures (in 204,515 patients) at surgical centers following at least 6 months of health plan enrollment. The study period began on January 1, 2005 and ran through December 31, 2012, with follow-up for endophthalmitis continuing for another 90 days through March 31, 2013. Cases were excluded if there was a prior diagnosis of endophthalmitis, if phacoemulsification was combined with corneal transplant or glaucoma surgery, if the planned procedure was complex, or if procedures were performed by retinal or oculoplastics specialists.

Data Collection

Postoperative infectious endophthalmitis was defined as occurring between the first post-operative day and 90 days after phacoemulsification. Preliminary cases of endophthalmitis were defined using diagnosis codes, or having an eye fluid specimen submitted to microbiology within 90 days of phacoemulsification. Endophthalmitis cases were later validated using detailed medical record review using a defined protocol which required a diagnosis recorded by a retina specialist within 90 days of surgery and evidence of intravitreal antibiotic therapy. We did not require microbiological confirmation, although positive cultures were noted. The review included the operative report, first follow-up visit, visits to retina specialists, and microbiology results.

Topical antibiotic orders and dispensings were obtained from the computerized pharmacy management information system, the gold standard, from which we obtained details regarding the following ophthalmic antibiotic preparations: gatifloxacin (Zymar or Zymaxid, Allergan, Parsippany, NJ), ofloxacin (Floxin, Daiichi Sankyo, Tokyo, Japan; Ocuflox, Allergan, Parsippany, NJ), polymyxin B/neomycin/dexamethasone (Maxitrol, Alcon, Hunenberg, Switzerland), polymyxin B sulfate/trimethoprim sulfate (Polytrim, Allergan, Parsippany, NJ), and gentamicin sulfate (various manufacturers) or tobramycin (Alcon, Hunenberg, Switzerland). We included dispensings recorded 90 days before phacoemulsification due to the timing of pre-operative visits, during which surgeons wrote the order for topical antibiotic to be filled in advance of the surgery. Agents dispensed up until the day after phacoemulsification were included to allow for late orders and dispensings while clearly separating prophylactic and therapeutic indications.

Intracameral injection was coded using batch processing of operative notes (N=315,246). Batch processing of intracameral injection information was performed without knowledge of endophthalmitis status to prevent investigator bias in adjudicating the cases. The positive and negative predictive values of the algorithm used to ascertain intracameral agent were 99.9% (95% CI, 99.4–100%). As the present case series is restricted to endophthalmitis cases, we allowed ourselves to re-examine the operative reports to confirm intracameral injection and agent. Two of the 215 cases previously classified as receiving an intracameral injection following batch processing for the prior study were reclassified as receiving topical agent only following chart review for this study.

Culture date, organism isolated, and antibiotic susceptibility testing results were obtained from clinical microbiology data. The majority of samples were obtained from the vitreous (N=94, 53%). All samples were sent for gram stain and culture. The culture media used was Trypticase™ soy agar with 5% sheep blood and chocolate agar (Becton Dickinson, Franklin Lakes, NJ). Antimicrobial susceptibility testing was performed in the laboratory using BBL™ Sensi-Discs™ (Becton Dickinson, Franklin Lakes, NJ) which are based on the Kirby-Bauer method. Interpretation of bacterial growth followed Clinical and Laboratory Standards Institute (CLSI) guidelines, or Sensi-Disc recommendations when CLSI guidelines were lacking (such as the interpretation of Staphylococcus species in the setting of bacitracin). Although sensitivity to moxifloxacin was not performed, sensitivity to gatifloxacin was considered a substitute for moxifloxacin sensitivity. Staphylococcus or Streptococcus species that are susceptible to gatifloxacin (or ciprofloxacin or ofloxacin) are considered susceptible to moxifloxacin as the latter has the most Gram-positive activity. Resistance to oxacillin was also deemed representative of clinical resistance to cephalosporins in Staphylococcus species.

Data Analysis

The approach to this descriptive analysis was informed by the work of Moloney and Park (2014).⁴ Crude counts are presented in relation to antibiotic prophylaxis and identified organism.

RESULTS

As described in our previous publication,¹² one-third of endophthalmitis cases were under 70 years of age, one-third were aged 70–79, and one-thid were 80 and older. 16% of cases had a history of diabetic retinopathy, and 4.2% experience intraoperative posterior capsular rupture. In addition 78% received topical antibiotic prophylaxis alone, 13% intracameral with or without topical prophylaxis, and 9% had no record of either approach. Among the 215 endophthalmitis cases, 38 (18%) were not cultured. Of the 177 cases that were cultured, 94 (53%) had no growth.

Of the 83 cases with organisms identified (47% of those cultured) (Table 1), 80 were Gram-positive. CoNS was the predominant organism in the series, accounting for 43 (52%) culture-confirmed cases.

Table 1.

Pathogen profile in relation to route of antibiotic prophylaxis (intracameral, topical).

	Intracameral (IC) antibiotic		No IC, topical antibiotic only			No IC or topical agent	Total
Organism isolated	With topical agent	Without topical agent	Topical ofloxacin	Topical gatifloxacin	Topical amino- glycoside
Gram-positive organisms
Coagulase-negative Staphylococcus	3	1	16†	12†	3	9	43
Streptococcus viridians	2	-	4	5	1	2	14
Methicillin-resistant S. aureus	-	-	2	1	1	3	7
Methicillin-sensitive S. aureus	-	-	2	2	-	1	5
Other Streptococcus species‡	-	-	1	2	-	2	5
Enterococcus species§	-	-	2	1	1	1	5
Propionibacterium acnes	1	-	-	-	-	-	1
Gram-negative organisms
Kingella species	-	-	-	1	-	-	1
Pseudomonas aeruginosa	-	-	-	-	-	1	1
Moraxella species	1	-	-	-	-	-	1
Total	7	1	27	24	6	19	83

^†One case, shown in two columns, received both topical ofloxacin and topical gatifloxacin.

^‡Three cases of S. pneumonia, one case of S. agalactiae (received topical gatifloxacin only) and one case of Streptococcus unspecified (topical ofloxacin only).

^§Four cases of E. faecalis and one case of Enterococcus species unspecified (received topical gentamicin only).

Among the Gram-positive organisms, information on sensitivity to any antibiotic was recorded for 68 (85%). Information on antibiotic sensitivity is summarized in Table 2. Among the 34 CoNS isolates with sensitivity testing, 58% (7 of 12 isolates) were sensitive to gatifloxacin, 57% (8 of 14) to ofloxacin and to polymyxin, 79% (11 of 14) to bacitracin and to trimethoprim, and 100% (N=32) to vancomycin. . Streptococcus viridans (13 of 14 isolates were tested) was 100% sensitive to cefazolin (N=4), ceftriaxone (N=9), cephalothin (N=3), ofloxacin (N=8), bacitracin (N=8), and clindamycin (N=6). Fifty percent (2 of 4 isolates) were sensitive to trimethoprim, 100% (N=10) to vancomycin, and 86% (6 of 7) to gatifloxacin. Other Streptococcus species (5 isolates, including 3 S. pneumoniae, 1 S. agalactiae, and 1 unspecified species) were 100% sensitive to ceftriaxone (N=2), bacitracin (N=2), cefazolin (N=1), cephalothin (N=1), clindamycin (N=4), and vancomycin (N=5). Fifty percent were sensitive to ofloxacin (1 of 2 isolates) and gatifloxacin (1 of 2)..

Table 2.

Isolates and antibiotic sensitivities in 68 Gram-positive culture-confirmed endophthalmitis cases following phacoemulsification surgery (number sensitive / number tested).

Antibiotic	Coagulase -negative Staphyloc occus N=34	Methicillin- resistant Staphyloc occus aureus N=6	Methicillin- sensitive Staphyloc occus aureus N=5	Streptoco ccus viridans N=13	Streptoco ccus species N=5	Enteroco ccus N=5
Cephalospor ins
Cefazolin	13/13	0/3	2/3	4/4	1/1	0/1§
Ceftazidime	6/17	0/3	0/1	1/2	1/1	0/1
Ceftriaxone	9/16	0/3	1/1	9/9	2/2	0/1
Cefuroxime	7/12	1/3	--	--	1/1	0/1
Cephalothin	15/15	3/3	1/1	3/3	1/1	0/1
Fluoroquinol ones †
Ciprofloxacin	11/18	0/4	1/1	0/3	1/1	1/3
Gatifloxacin (surrogate for moxifloxacin)	7/12	1/3	2/2	6/7	1/2	2/3
Ofloxacin	8/14	0/3	2/3	8/8	1/2	0/1
Aminoglycos ides
Gentamicin	19/19	4/4	3/3	3/5	0/1	0/1§
Amikacin	14/15	1/3	1/1	0/4	0/1	0/2§
Neomycin	13/14	0/3	3/3	0/8	0/2	0/2§
Tobramycin	5/5	0/1	1/1	1/1	--	--
Polypeptides
Bacitracin	11/14	1/3	2/3	8/8	2/2	1/2
Polymyxin B	8/14	0/3	0/3	0/8	0/2	0/2
Other
Clindamycin	27/32	0/6	2/3	6/6	4/4	0/1§
Oxacillin‡	18/30	0/6	3/3	2/5	2/3	0/1
Sulfisoxazole	11/14	2/3	2/3	1/3	1/1	0/1
Tetracycline	20/22	3/3	3/3	1/2	0/1	--
Trimethoprim	11/14	3/3	3/3	2/4	0/1	1/1§
Vancomycin	32/32	6/6	5/5	9/10	5/5	4/4

^†Staphylococcus or streptococcus species which are susceptible to gatifloxacin (or ciprofloxacin or ofloxacin) are considered susceptible to moxifloxacin because moxifloxacin has the greatest Gram-positive activity.

^‡If a Staphylococcus organism is sensitive to oxacillin, it is considered sensitive to all the cephalosporins. If it is resistant to oxacillin, it is considered resistant to all the cephalosporins.

^§Enterococcus is considered intrinsically resistant to cephalosporins, clindamycin and trimethoprim. Organisms tested as sensitive to trimethoprim may have in vitro activity, but trimethoprim is not clinically effective against Enterococcus. In addition, aminoglycosides are not effective against Enterococcus even if in vitro testing suggests susceptibility; higher doses would be required and would need to be specifically tested.

All Staphylococcus aureus isolates (MRSA, N = 7; MSSA, N=5) were susceptible to vancomycin. Only 33% (1 of 3) of MRSA isolates were sensitive to gatifloxacin or bacitracin, while 0% (N=3) were sensitive to ofloxacin or polymyxin. In addition, 100% (N=3) were sensitive to trimethoprim. MSSA isolates were 100% sensitive to gatifloxacin (N=2) and trimethoprim (N=3). Sixty-seven percent of MSSA isolates (2 of 3 isolates) showed sensitivity to ofloxacin or bacitracin. All Streptococcus isolates (Streptococcus viridans, Streptococcus aureus, and other Streptococcus species) were resistant to polymyxin.

Enterococcus (5 isolates) was 100% sensitive to vancomycin. Enterococcus is intrinsically resistant to cephalosporins, clindamycin, trimethoprim, and aminoglycosides. Three isolates were tested for sensitivity to gatifloxacin (a moxifloxacin surrogate), and 1 was resistant.

Among the culture-confirmed cases where a topical antibiotic was the sole prophylactic agent, 45 had sensitivities measured to one or more antibiotics that are relevant to cataract surgery (Table 3). Among 10 isolates where topical ofloxacin had been prescribed and the organism was tested, 5 (50%) were sensitive to ofloxacin. Among 5 isolates where topical gatifloxacin had been prescribed and the organism was tested, 3 (60%) were sensitive to gatifloxacin. Among isolates where topical aminoglycoside (tobramycin or gentamicin) had been prescribed and the organism tested, 100% were sensitive to tobramycin and gentamicin. In summary, of 19 cases that had received topical antibiotic prophylaxis and had sensitivity testing performed, 7 (37%) were resistant to the antibiotic given.

Table 3.

Isolates and antibiotic sensitivities in 45 culture-confirmed, Gram-positive endophthalmitis cases following phacoemulsification surgery with only prophylactic topical antibiotics (number sensitive / number tested).^†

	Topical ofloxacin‡ (N=21)	Topical gatifloxacin‡ (N=18)	Topical aminoglycoside (N=6)
Oxacillin	10/18	9/15	3/5
Fluoroquinolones
Ciprofloxacin	5/9	2/7	2/5
Gatifloxacin	8/9	3/5	0/1
Ofloxacin	5/10	5/6	1/3
Aminoglycosides
Tobramycin	2/2	2/2	2/2
Gentamicin	11/11	7/9	4/4

^†The table does not show cases whose antibiotic sensitivity was not tested or was tested to other antibiotics.

^‡One case, shown in two columns, received both topical ofloxacin and topical gatifloxacin.

Pathogens were identified in only eight cases of endophthalmitis following intracameral injection, and only 3 cases had pertinent antibiotic sensitivity data. Two cases received intracameral moxifloxacin and topical gatifloxacin, of which one was associated with culture of CoNS (resistant) and the other S. viridans (sensitive). The third case, also S. viridans, received intracameral cefuroxime with topical gatifloxacin, to which the pathogen was sensitive. All cases who received intracameral antibiotic were sensitive to vancomycin. Three cases of endophthalmitis were associated with culture of a Gram-negative organism (Table 1). Kingella species was recovered in one case where topical gatifloxacin was the sole prophylactic agent; this case was not tested for antibiotic sensitivities. Pseudomonas aeruginosa was cultured from an eye that received no antibiotic prophylaxis and was sensitive to all agents tested (ceftazidime, ciprofloxacin, gentamicin). Moraxella species was recovered from an eye that had been injected with intracameral cefuroxime and supplemented with topical ofloxacin but was not tested for sensitivities. No fungal organisms were recovered despite routine plating of specimens on Sabouraud’s medium.

DISCUSSION

We assessed infective organisms and antibiotic sensitivities in 215 endophthalmitis cases relative to the route of administration and antibiotic agent used for prophylaxis. Consistent with other reports, about half of the cases submitted for microbiological testing were confirmed with an infective organism.¹ Among those that were culture-confirmed, 96% were Gram-positive, with Staphylococcus species (including MRSA and MSSA) accounting for 66% of organisms and MRSA itself making up 8% of isolates. Our finding that only 4% of isolates were Gram-negative corresponds to previous reports.^1,3,4 Sensitivity results from our study correspond to prior literature showing that both MRSA and MSSA have greater in vitro sensitivity to gentamicin than fluoroquinolones.^22,23 Our results also support prior reports that Streptococcus species is highly susceptible to ceftriaxone and vancomycin,²⁴ and that Enterococcus faecalis is highly sensitive to vancomycin.²⁵

Postoperative, culture-confirmed endophthalmitis can develop despite antibiotic prophylaxis for several reasons: (1) delayed wound healing;²⁶ (2) lack of sufficient penetration or dosing of a topical agent; and (3) antibiotic resistance. Corneal wounds in phacoemulsification are subject to incompetence, allowing fluid to enter the eye.^27,28 It is logical to expect that the probability of infection is increased if an organism enters through the wound in the hours or days following a decline in antibiotic concentration below the organism’s minimum inhibitory concentration (MIC). Topical agents reach relatively low levels of concentration in the anterior chamber^17,18,29,30 and are subject to peaks and troughs. Therefore, in addition to the practice of intracameral injection, wound construction and stromal hydration are important in the goal to maintain a watertight wound.³¹

Our earlier report described an important difference in effectiveness between topical aminoglycosides (tobramycin, gentamycin) and topical fluoroquinolones (gatifloxacin and ofloxacin).²⁰ Although aminoglycosides are highly effective against Gram-positive and Gram-negative species in vitro,³ recent animal studies indicate poor penetration into the eye’s anterior chamber.^23,29,32,33 Our findings are consistent with prior animal evidence: among 6 endophthalmitis isolates that had received topical aminoglycoside prophylaxis and were tested, 100% were sensitive. In contrast, only 50% of 10 isolates that had received topical ofloxacin prophylaxis were sensitive to ofloxacin, while 60% of 5 isolates that had received topical gatifloxacin prophylaxis were sensitive to gatifloxacin. Resistance to fluoroquinolones has been confirmed in other studies^{3,22,23,34,35} and calls into question the recommended usage of these antibiotics for routine prophylaxis.

We have separately reported the effectiveness of intracameral injection, compared with topical antibiotic, for preventing endophthalmitis.¹² We found the use of intracameral antibiotics to be associated with a significantly lower rate of endophthalmitis following cataract surgery (intracameral compared with topical: odds ratio 0.58 with 95% confidence interval 0.38–0.91). This result is similar to other reports.^1,6–9 This supports the importance of administering a high concentration (compared to MIC90 of most causative organisms) of antibiotic directly into the anterior chamber.^{36, 37} Studies suggest that topical antibiotics may not add any additional clinical benefit to intracameral injection.^9,19,21

Vancomycin would appear to be an excellent choice for intracameral injection given that over 96% of organisms recovered in our setting were sensitive to that agent. Tempering the enthusiasm for vancomycin are prior recommendations from the CDC reserving vancomycin for non-routine infections and recent reports linking intracameral vancomycin to hemorrhagic occlusive retinal vasculitis.^{38, 39} Although exceedingly rare, the condition most commonly results in a devastating loss of vision. This latter concern prompted the American Society of Cataract and Refractive Surgery to issue an advisory warning.⁴⁰ Intracameral cefuroxime and moxifloxacin are similar to each other in reducing endophthalmitis rates.²¹. Although neither of these agents are available as government approved, manufactured products in the United States, cefuroxime is available in the European Union as Aprokam (Thea pharmaceuticals, Keele, UK) and moxifloxacin is available in India as Promox ( Aurolab, Tamil, Nadu, India).

Other methods to increase the concentration of antibiotic in the anterior chamber include improved formulations of topical agents and bottles (such as medication concentration, pH, and instillation volume) that could improve corneal penetration, particularly in regards to aminoglycosides.⁴¹ Intraocular lenses soaked in antibiotics and antibiotic nanoparticles may also hold a future role in endophthalmitis prevention.^42,43

Although this study accessed endophthlamitis cases from a large population, 18% of the cases were not cultured. This is a consequence of the study being set in a community-based population. The setting was both a limitation and a strength, in that it represents real-world care in a diverse population. The cases were obtained as a consecutive case series from a well-characterized population and represent the distribution of organisms in our underlying geographic region. Nonetheless, caution must be exercised in applying these findings to other geographic regions, where the distribution of pathogens and their sensitivities may differ.

In summary, Gram-positive organisms, and CoNS in particular, were the predominant organisms associated with endophthalmitis in this series. More than 50% of CoNS showed resistance to fluoroquinolones. S. viridans, which accounted for the next most prevalent organism, was predominantly sensitive to fluoroquinolones. Topically applied fluoroquinolones have been associated with increased likelihood of emergence of antibiotic resistance.^44,45 While pathogens were generally susceptible to aminoglycosides in vitro, our previous report suggests that these are not effective in endophthalmitis prophylaxis, most likely due to poor ocular penetration.