Introduction

Conjunctivitis is the most common ocular disease worldwide [1, 2]. Bacterial conjunctivitis, the predominant kind of conjunctivitis, has the potential for significant ocular morbidity. Rapid destruction of the eye is enhanced by the presence of purulence, which characteristically accompanies bacterial infection [2–4].

Although most cases of conjunctivitis are self-limited, treatment with antibiotics has been shown to decrease discomfort and duration of symptoms, as well as diminish the potential for morbidity [5–8]. Frequently, antibiotics are started empirically before culture results are available, in order to increase the efficacy of treatment.

The ideal topical antibiotic for empirical treatment of bacterial conjunctivitis is the one that exhibits broad-spectrum coverage for ocular pathogens, including gram-positive and gram-negative bacteria [9]. Periodic studies to monitor the emerging antibiotic resistance trends are, therefore, crucial in guiding antibiotic selection.

This study was designed to determine the most common pathogens responsible for bacterial conjunctivitis and their in vitro susceptibility to existing antibiotics, as well as the changing trends in the resistance of these bacteria to antibiotics over the last decade. To the best of our knowledge, this is the largest study of its kind to date, and also the first of its kind in the northeastern United States.

Materials and methods

After obtaining New York Eye and Ear Infirmary (NYEEI) IRB ethics committee approval, microbiology records for all conjunctival cultures performed from 1 January 1997 through 30 June 2008 for presumed bacterial infections were reviewed. Culture samples were obtained following standard NYEEI protocol aseptic technique, according to which the specimen was collected with a cotton or Dacron (DuPont, Wilmington, DE, USA) swab under sterile conditions. Culture and sensitivity testing immediately followed specimen collection. The collections were inoculated onto chocolate agar, trypticase soy agar with 5% sheep blood, and thioglycollate broth (Becton Dickinson, Franklin Lakes, NJ, USA) and then placed in a 5% to 7% CO2 incubator at 37 degrees Celsius for 24 to 48 hours. Cultures were considered positive if they demonstrated heavy growth on media or thioglycollate, and/or moderate growth on chocolate agar, followed by identification of bacterial pathogen. The cultures were excluded from review if the identified organisms were constituents of normal conjunctival flora. The exception was made for cultures which grew Streptococcus viridans (S. viridans), which despite its being a conjunctival commensal has been shown to be pathogenic in ophthalmic infections [9, 10]. Cultures that grew fungus only, or fungus in conjunction with bacteria, were excluded.

The National Committee for Laboratory Standards guidelines were followed for routine disc diffusion (Kirby–Bauer method) and/or microdilution (Vitek 2, Automatic Microbial System, Raleigh, NC, USA) for minimum inhibitory concentration (MIC) susceptibility testing for the antibiotics tetracycline, trimethaprim/sulfa, imipenem, fluoroquinolones (ciprofloxacin, moxifloxacin, gatifloxacin), aminoglycosides (gentamicin, tobramycin), erythromycin, cefazolin, oxacillin, and vancomycin. MIC for some organisms (i.e., S. viridans, H. influenzae, S. pneumoniae, Moraxella sp., and Nisseria sp.) cannot be done using Vitek 2; thus, disc diffusion was utilized.

Statistical analysis was performed using a linear regression model to estimate the average biennial or annual change in the percentage of bacterial isolates, antibiotic resistances and their 95% confidence intervals (CI), as well as to test the statistical significances of time trends.

Results

The NYEEI microbiology laboratory performed 20,180 cultures for presumed bacterial conjunctivitis between 1 January 1997 and 30 June 2008. Of these, 12,134 (60.1%) were considered to be culture-positive, and 39.9% were considered culture-negative (21.5% grew conjunctival commensals other than S. viridans, and 18.4% had no growth). Fungi were identified in 0.8% of all cultures.

Of the culture-positive isolates, 9,920 (76.6%) were gram-positive and 2,844 (23.4%) gram-negative pathogens (Table 1). The percentage of gram-positive isolates increased with an average annual change (AAC) of +2.10% on an estimated linear regression model (p = 0.0041) over 11.5 years. Gram-negative isolates decreased with an AAC of −2.10% (p = 0.0042) over the same period.

Table 1 Positive cultures from 1997 to 2008
Full size table

Table 2 and Fig. 1 are representations of the prevalence of the six most commonly encountered isolates. Staphylococcus aureus (S. aureus) was the most commonly identified isolate (38.7%), and Serratia marcescens (S. marcescens) was the least prevalent isolate (2.4%) of the group. S. aureus constituted 50.6% of all gram-positive isolates [methicillin-resistant S. aureus (MRSA) = 15.2%, methicillin-sensitive S. aureus (MSSA) = 35.4%] , followed by S. viridans (10.5%) and Streptococcus pneumoniae (S. pneumoniae) (10.05%). Haemophilus influenzae (H. influenza) was the most common gram-negative isolate (29.3%), followed by Pseudomonas aeruginosa (P. aeruginosa) (20.5%), and Serratia marcescens (S. marcescens) (10.2%).

Table 2 Prevalence of the top six isolates
Full size table
Fig. 1
figure 1

Prevalence of top six bacterial isolates

Full size image

Table 3 summarizes the average biennial change (ABC) for the top six isolates in the course of 11.5 years. S. aureus demonstrated an average biennial increase of 2.78% (p = 0.011), as compared to other gram-positive isolates. Steady ABC in percentages of isolates were also noted for S. pneumoniae (+0.87%; p = 0.005), S. viridans (+1.55%; p = 0.007%), and P. aeruginosa (+1.94%; p = 0.009). In contrast, H. influenzae and S. marcescens decreased in frequency over the years, although the trends were not statistically significant (−1.51%; p = 0.186 and −0.81%; p = 0.260 respectively).

Table 3 Average biennial change (ABC) and 95% confidence interval (CI) for the top six isolates, from 1997–1998 to 2007–2008
Full size table

Table 4 summarizes the average annual change (AAC) of bacterial resistances to tetracycline, trimethaprim/sulfamethoxazole (TMP/SMZ), ciprofloxacin, gentamicin, tobramycin, erythromycin, cefazolin, and oxacillin.

Table 4 Average annual change (AAC) of antibiotic resistances from 1997 to 2008
Full size table

Tobramycin was tested on all 12,134 isolates. Figure 2 illustrates the average annual change (AAC) of tobramycin resistance over the 11.5 years. Although there was a decrease in the resistance of the entire gram-positive isolates group to tobramycin (AAC = −0.96%; p = 0.0012), S. aureus resistance to the antibiotic increased (AAC = +0.36%; p = 0.0106). Gram-negative isolates similarly showed an increase in resistance (AAC = +0.40%; p = 0.127) with the exception of H. influenzae, which displayed a constant low resistance rate of 1–2%.

Fig. 2
figure 2

Trends in resistance of bacterial isolates to the aminoglycosides

Full size image

Gentamicin was also tested on all 12,134 isolates. Figure 2 illustrates the AAC of gentamicin resistance over the 11.5 years. While gram-positive isolates as a group demonstrated decrease in resistance to the antibiotic (AAC = −0.57%, p = 0.061), S. aureus resistance to gentamicin increased (AAC = +0.36%, p = 0.0106). The gram-negative isolates showed an increased resistance to gentamicin, (AAC = +0.36%, p = 0.171), with the exception of H. influenzae, which demonstrated a constant low resistance rate of 1–2%.

Erythromycin was tested on gram-positive isolates only (9,290 cultures). Figure 3 shows erythromycin resistance over the 11.5 years. There was a steady increase in resistance to erythromycin of gram-positive isolates (AAC = +1.66%; p < 0.0001). This trend was particularly significant in S. aureus (AAC = +3.74%; p < 0.0001).

Fig. 3
figure 3

Trends in resistance of bacterial isolates to erythromycin

Full size image

Ciprofloxacin was tested on all 12,134 isolates. Figure 4 shows ciprofloxacin resistance over the 11.5 years. There was a steady increase in ciprofloxacin resistance of the entire gram-positive isolates group (AAC = +1.60%; p = 0.0020), with S. aureus again showing the most significant increase (AAC = +2.57%; p = 0.0005). Resistance of the entire group of gram-negative isolates increased also (AAC = +1.51%; p = 0.0131). H. influenzae and alpha streptococci were exceptions, demonstrating a steady low resistance less than 8% throughout the 11.5 years.

Fig. 4
figure 4

Trends in resistance of bacterial isolates to ciprofloxacin

Full size image

Testing for gatifloxacin and moxifloxacin resistance commenced in 2003 for gram-positive and gram-negative isolates, and was performed on 6,801 isolates. Figure 5 shows gatifloxacin and moxifloxacin resistance over the 5 years. For the gram-positive pathogens, resistance stayed constant at 3–5% between 2003 and 2008, then abruptly increased to15% in 2008. A more gradual increase in resistance was observed for gram-negative isolates (3–6% between 2003 and 2008). S. aureus resistance to gatifloxacin and moxifloxacin increased from 0% to 4% and from 0% to 5% respectively, between 2003 and 2008.

Fig. 5
figure 5

Trends in resistance of bacterial isolates to the fourth generation fluoroquinolones

Full size image

Trimethoprim/sulfamethoxazole (TMP/SMZ) was tested on all 12,134 isolates. Figure 6 shows TMP/SMZ resistance over the 11.5 years. There was an increase in resistance of gram-positive (AAC = +2.34%; p < 0.0001) and gram-negative (AAC = +3.12%; p < 0.0001) isolates to TMP/SMZ, particularly notable in S. pneumoniae, H. influenzae, and S. aureus (AAC = +2.66%, p < 0.0001; AAC = +1.39%, p < 0.0001; AAC = +1.23%, p < 0.0001 respectively). Only S. viridans showed a decrease in resistance to TMP/SMZ, but the finding was not statistically significant (AAC = −0.26%, p-value = 0.5374).

Fig. 6
figure 6

Trends in resistance of bacterial isolates to trimethoprim/sulfamethoxazole

Full size image

Tetracycline was tested on all 12,134 isolates. Figure 7 shows tetracycline resistance over the 11.5 years. There was an increase in resistance of all gram-positive (AAC = +5.01%; p < 0.0001) and gram-negative (AAC = +7.41%; p < 0.0001) isolates to tetracycline, most prominent in H. influenzae and S. viridans (AAC = +2.18%, p < o.0001; AAC = +2.02%, p < 0.0001 respectively). S. aureus showed a statistically insignificant overall increase in tetracycline resistance in 11.5 years (AAC = +1.42, p = 0.3731). This organism displayed a bimodal resistance pattern, however, with rapid increase from 5% to 62% between 1999 and 2000, followed by a more gradual decline in resistance to 22% in 2008.

Fig. 7
figure 7

Trends in resistance of bacterial isolates to tetracycline

Full size image

At NYEE, imipenem is tested against gram-negative pathogens only (2,844 isolates). Imipenem resistance remained 0%–1% over the course of 11.5 years (Fig. 8).

Fig. 8
figure 8

Trends in resistance of bacterial isolates to imipenem

Full size image

Cefazolin was tested on all 12,134 isolates. Figure 9 shows cefazolin resistance over the course of 11.5 years. There was an increase in resistance of both gram-positive and gram-negative isolates to this antibiotic (AAC = +0.68%; p = 0.2043 and AAC = +0.84%; p < 0.0001 respectively). S. aureus showed greatest increase in cefazolin resistance (AAC = +4.02%; p < 0.0001).

Fig. 9
figure 9

Trends in resistance of bacterial isolates to cefazolin

Full size image

Oxacillin was tested against gram-positive isolates only. Figure 10 shows oxacillin resistance over the course of 11.5 years. There was a steady increase in oxacillin resistance of all gram-positive pathogens (AAC = +4.39%; p < 0.0001). Methicillin-resistant S. aureus (MRSA) was found in 2,632 of 9,290 isolates (28.3%). The frequency of MRSA isolates steadily increased throughout the 11.5 years (ACC = +3.69%; p < 0.0001).

Fig. 10
figure 10

Trends in resistance of bacterial isolates to oxacillin

Full size image

Vancomycin resistance was tested on gram-positive pathogens only; they were 100% sensitive to the antibiotic.

Discussion

We observed several significant trends in the prevalence of conjunctival bacterial isolates and in their resistance to various antibiotics over the past decade at the New York Eye and Ear Infirmary. The ratio of gram-positive to gram-negative isolates increased significantly (p = 0.004). High frequencies of S. aureus, S. viridans, S. pneumoniae, H. influenzae, P. aeruginosa, and S. marcescens were noted, similar in finding to previously reported studies [11–13].

Gram-positive isolates group demonstrated greater resistance to aminoglycosides than the gram-negative isolates in the late 1990s. The resistance of gram-positive isolates group to these antibiotics decreased significantly in the course of 10 years, whereas gram-negative organisms demonstrated an increase in resistance, albeit statistically insignificant. These trends are comparable to the previously reported studies on antibiotic susceptibilities of ocular isolates, and probably reflect increased frequency of aminoglycosides selection for presumed and culture-proven gram-negative infections [11, 12, 14].

Despite a steady increase in resistance of gram-negative pathogens to aminoglycosides, data from 2008 show that resistance rate is still relatively low. Overall, 12% of gram-negative isolates were resistant to aminoglycosides in the past year, and H. influenzae resistance was as low as 2%. Similarly, 12% of the gram-positive isolates group were resistant to tobramycin and 20% to gentamicin. It is noteworthy that S. aureus showed relatively low resistance rates to tobramycin (10%) and gentamicin (12%). These findings suggest that both aminoglycosides are still reasonably good treatment options for bacterial conjunctivitis.

Erythromycin, the only macrolide in the study, was tested only on gram-positive isolates. There was an overall 2-fold increase in resistance (24–45%) to erythromycin over 11.5 years, with the most significant increases observed in S. aureus (AAC = +3.74%; p < 0.0001) and alpha-Streptococci (AAC = +3.23%; p < 0.0001). These findings are similar to those reported by Cavuoto et al. [12]. Data from 2008 demonstrated that nearly a half of gram-positive isolates were resistant to erythromycin, suggesting that this medication is currently not an ideal choice for empiric antibiotic coverage.

Fluoroquinolones are frequently used for treating conjunctivitis and other ocular infections. We observed a 6-fold increase in resistance of the gram-positive isolates group to ciprofloxacin (5–30%; AAC = +1.60%; p = 0.002), while the gram-negative isolates group had a less significant increase in resistance (1–16%; AAC = +1.51%; p = 0.0131). Streptococci and H. influenzae maintained a consistently low level of ciprofloxacin resistance (2–6%). All isolates demonstrated low resistance to gatifloxacin and moxifloxacin (0–6%) until the last year, during which a 4- to 5-fold increase in resistance of the gram-positive isolates to the latest generation of quinolones was observed. With the exception of imipenem, moxifloxacin and gatifloxacin had the lowest resistance rates among gram-negative isolates. The two aforementioned antibiotics also have the lowest resistance rates for gram-positive isolates after vancomycin. Of note, S. aureus demonstrated low resistance rates to gatifloxacin and moxiflocxacin (4% and 5% respectively). These findings are similar to those of Jensen et al. [9]. The observation of uniformly low resistance makes 4th-generation quinolones the best choice for empiric broad-spectrum antibiotic coverage. On the other hand, a 4-fold increase in the resistance of gram-positive isolates to these antibiotics in the past year may signal the beginning of an increasing resistance trend.

There was an overall 5-fold increase in the resistance of gram-positive isolates, and 3.5-fold increase in the resistance of gram-negative pathogens to TMP/SMZ. Of all isolates, S. pneumoniae showed the greatest increase at AAC (+2.66%). The increase in resistance in the recent years is greater than values reported by Cavuoto et al. [12]. The discrepancy may reflect inclusion of more recent data in our study. This is supported by the fact that the resistance rates were comparable in the overlapping years of both studies. Additionally, the difference in resistance rates could be attributable to the disparity in the geographic setting of the two studies.

Tetracycline showed an overall 6-fold increase in resistance of gram-positive isolates, and a 15-fold increase in resistance of gram-negative isolates. The data from recent years showed resistance levels up to 90% for gram-negative isolates, a trend that could explain why tetracycline has become an unpopular antibiotic for treatment of bacterial conjunctivitis [15]. Notably, it continues to be effective against S. pneumoniae (10% resistance in 2008). S. aureus’ resistance to tetracycline gradually declined from 60% in 2000 to 20% in 2008, likely reflecting tetracycline’s reduced use [15].

Imipenem, tested only against gram-negative pathogens in this study, showed a negligible resistance of 0–1% throughout the years. This low resistance is probably due to the fact that imipenem is not currently commercially formulated for ophthalmic use. Imipenem has the potential of becoming an effective ocular antibiotic in the future.

There was a 1.5-fold increase in gram-positive isolate resistance and about 3-fold increase in gram-negative pathogen resistance to cefazolin. S. aureus showed a most prominent increase in resistance, similar to the observations of Chalita at al. [11]. Gram-positive pathogens demonstrated higher levels of cefazolin resistance than gram-negative isolates, probably due to the fact that cefazolin is used more often for suspected and culture-proven gram-positive infections [16].

Oxacillin, a penicillinase-resistant antibiotic often reserved for detection of MRSA, exhibited an increase in resistance of 2%–40% for S. aureus, reflecting a significant increase in prevalence of MRSA at our institution. This finding is also similar to those reported in previous studies [12, 17–19].

There was no resistance of any gram-positive pathogens to vancomycin, which makes it an ideal antibiotic for MRSA. It is advisable to limit vancomycin use to MRSA at the present time to minimize the risk of resistance [20].

It is to be acknowledged that the in vitro susceptibility tests in this study were based on serum antibiotic concentrations, and may not reflect the efficacy of the antibiotic in the eye [21, 22]. It is, therefore, possible that in vivo resistances are actually different from what we have demonstrated in in vitro assays. It must also be noted that this study only includes data from the New York Eye and Ear Infirmary laboratory, and that this does not represent a regional or national trend.

With the treatment of acute bacterial conjunctivitis, one should be reminded that most cases resolve spontaneously. While clinical remission is significantly observed early in the course of treatment (days 2–5), the benefit decreases to marginal for later remission (days 6–10) [23].

In summary, this study outlines the prevalence trends in most common conjunctival bacterial isolates and the trends in their resistance rates to the selected antibiotics in the Northeast United States over the past decade. Our study demonstrated that gram-positive organisms have become more frequently identified as etiologic agents of bacterial conjunctivitis. The conjunctival bacterial isolates showed high levels of resistance against tetracycline, erythromycin, and TMP/SMZ, suggesting that these antibiotics are not the ideal choices for empirical broad-spectrum coverage. Moxifloxacin and gatifloxacin currently offer the best broad-spectrum coverage, and aminoglycosides are the second best option. However, a marked increase in resistance to fluoroquinolones in the recent years prompts cautious use of these antibiotics. Vancomycin continues to be the best antibiotic for MRSA coverage.