Background

The eye, a functionally and structurally complex organ, experiences a variety of bacterial, viral, fungal and parasitic infections [1–3]. Bacterial infections (Both Gram-positive and Gram-negative) contribute to 32 to 74% of ocular infections globally [4–10]. Staphylococci are the leading ocular isolates worldwide among the Gram-positive bacteria, [9, 11–13]; while Pseudomonas aeruginosa, Klebsiella pneumoniae and Escherichia coli are the major Gram-negative bacteria isolated from ocular infections [3, 14, 15].

Bacteria are the most common agents causing external ocular infections, including blepharitis, keratitis, dacryocystitis, and orbital cellulitis. Bacteria are responsible for 70–80% of conjunctival morbidity which poses a huge socioeconomic burdens to the general public [16–19]. Blepharitis, an inflammation of the eyelid, can cause complete loss of the eyelashes if not diagnosed early [20]. Bacterial keratitis is the leading cause of corneal blindness [18, 21, 22].

External ocular infections may remain localized, or progress to adjacent tissues. For instance, external and internal hordeolum can result from the spread of eyelid infection to the respective glands [23]. Dacryocystitis, inflammation of the nasolacrimal duct, is the result of nasolacrimal duct obstruction. The accumulation of fluid and edema in the eye secondary to dacryocystitis, is a potential danger to ocular tissues such as the cornea and conjunctiva [24, 25]. External ocular infections that result from systemic spread, surgery or are secondary to ocular trauma, can lead to sight threatening intraocular infections such as endophthalmitis [1, 22, 26, 27]. Left untreated, bacterial ocular infections, bacterial products and vigorous inflammation following infection, can irreversibly damage ocular structures [28, 29]. Inflammation and scarring, once present, may not be easily resolved and can result in visual impairment or permanent loss of vision [30, 31].

Additionally, bacterial ocular infections have been complicated by multidrug resistance;, a problem that is intensifying over time [32, 33]. This poses a challenge in the clinical management of bacterial ocular infections [34–36].

Eye infections are remarkably common in northern Ethiopia; however, no studies have been conducted on this topic in this region. Similarly, the antimicrobial susceptibility patterns of infecting bacteria in this region are not known. Therefore, we sought to determine the ocular bacterial infection, the risk factors and the antimicrobial susceptibility of bacterial isolates in ocular infections at Quiha Ophthalmic Hospital, Tigray, northern Ethiopia.

Methods

Study design, data and specimens collection

A hospital based cross-sectional study was conducted from September-December 2015 in Quiha Ophthalmic Hospital, Quiha, Tigray, Ethiopia. A total of 270 patients who presented with suspected ocular infection were consecutively included in the study. Sociodemographic, clinical and risk-factor related data were collected using a pretested structured questionnaire using interviewer-administered questionnaires. Study subjects were examined using a slit-lamp biomicroscope to screen for the presence of ocular infection. Ocular specimens were collected from periorbital cellulitis (N = 8), dacryocystitis (N = 13), blepharitis (N = 26), blepharoconjunctivitis (N = 16), conjunctivitis (N = 123), keratitis (N = 57) and endophthalmitis (N = 27) diagnoses. An ophthalmic surgeon collected specimens from all patients except those patients who had endophthalmitis who had their specimemens collected by an ophthalmologist [37–39].

Specimen transportation

The respective specimens were then transferred into 2 ml of brain-heart infusion broth (Oxoid, Hampshire, UK). Tubes were tightly capped, gently mixed, labeled and placed in cold chain. Finally, samples were transported to the Microbiology Laboratory, Mekelle University, for microbiological analysis [38, 39].

Cultivation and identification of isolates

The broth was gently mixed to homogeneity; 100 μl of inoculum was then dispensed and streaked onto blood agar, chocolate agar, MacConkey agar, Mannitol salt agar and modified Thayer-Martin agar (MTM) (Oxoid, Hampshire, UK). To maintain the presence of CO2, chocolate agar and MTM agar plates were placed in candle jars. All media were incubated overnight at 37 °C. After 24 h of incubation, each plate was inspected for any growth and negative plates were incubated for an additional 24 h. For eyelid and conjunctival swabs, culture positivity was determined based on a threshold criteria. Corneal specimens were considered as positive if there was a confluent growth at the site of inoculation [39]. All bacterial isolates were identified using standard clinical laboratory methods [40, 41].

Antimicrobial susceptibility testing

Antimicrobial susceptibility testing was performed using the standard Kirby-Bauer disk diffusion method [41]. For fastidious organisms, Muller-Hinton agar with 5% sheep blood was used and incubated in 5% CO2; GC agar and Haemophilus Test Medium were used for Neisseria gonorrhoeae and Haemophilus influenzae, respectively [42]. The following antibiotics were used: trimethoprim-sulfamethoxazole (SXT; 1.25/23.75 μg), erythromycin (15 μg), clarithromycin (15 μg), chloramphenicol (30 μg), clindamycin (2 μg), tetracycline (30 μg), doxycycline (30 μg), amikacin (30 μg), gentamicin (10 μg), ciprofloxacin (5 μg), ceftriaxone (30 μg) and cefoxitin (30 μg) (HIMEDIA). The zones of inhibition of the antimicrobial agents were measured using calipers; bacteria were described as susceptible or resistant based on CLSI guideline [42]. Multidrug resistance was defined as non-susceptible to ≥1 agent in ≥3 antimicrobial categories according to the definition of Magiorakos and colleagues [43].

Quality assurance

Standard operational procedures were strictly followed from the pre-analytical to the post-analytical phase. The questionnaire was pretested on 16 patients with ocular infection. Data collectors were trained on data collection procedures. Completed questionnaires were proofread immediately after completion by the research team to clarify any labelling errors or illegibility. Standard Operating procedures were followed during Specimen collection, handling, transportation, microbiological analysis and interpretation of results. Reagents, media and antimicrobial disks were checked for the expiry date, damage and storage problems. Media sterility was strictly monitored by random sampling and incubation to test for any growth. In addition, reference strains of Staphylococcus aureus ATCC25923, E. coli ATCC25922 and P. aeruginosa ATCC27853 were used to monitor the quality of this work [42].

Statistical analysis

Data was entered into SPSS version 21, and descriptive statistics, bivariate and multivariate logistic regression analyses were performed. Bivariate logistic regression was employed to look for association between the outcome variable and independent variables. Variables with P-values <0.05 were reanalyzed in multivariate logistic regression to identify risk factors for bacterial occurrence. The corresponding P-value <0.05 and the confidence interval (CI) of 95% were considered statistically significant.

Results

Sociodemographic characteristics

A total of 270 study subjects with clinically diagnosed ocular infections were included in the study, of whom 151 (55.9%) females, 119 (44.1%) were males and 158 (58.5%) were rural dwellers (Table 1). The mean (SD) age of the study subjects was 37.8 (22.9) years.

Table 1 Sociodemographic characteristics of the study subjects (N = 270)
Full size table

Bacterial profile and clinical features

Conjunctivitis (123 cases of 270, 45.6%) was the leading clinical presentation, followed by keratitis (57, 21.1%), endophthalmitis (27, 10.0%), blepharitis (26, 9.6%), blepharoconjunctivitis (16, 5.9%), dacryocystitis (13, 4.8%), and periorbital cellulitis (8, 3.0%) (Table 2). Of the total 270 patients, 180 (66.7%) were culture positive for bacterial isolates. Single and mixed bacterial infection patients were seen in 174 (96.7%) and 6 (3.3%) cases, respectively (Table 2). The mixed isolates were detected from three patients with conjunctivitis, one patient with periorbital cellulitis, one with keratitis and one with endophthalmitis diagnosis.

Table 2 The distribution of culture positive samples in ocular infection
Full size table

Bacterial infection was significantly higher in patients with dacryocystitis 92.3% (COR = 20, 95% CI: 2.0–195, P = 0.01), periorbital cellulitis 87.5% (COR = 11.7, 95% CI: 1.1–183.8, P = 0.03), keratitis 80.7% (COR = 6.9, 95% CI: 2.0–23.3, P < 0.01) and endophthalmitis 74.1% (COR = 4.7, 95% CI: 1.2–17.9, P = 0.02) than for the other diagnoses (Table 2).

Identification of bacterial isolates

In this study, 113 (60.7%) bacterial isolates were Gram-positive, and 87 (39.3%) were Gram-negative. Overall, the dominant bacterial isolates were S. aureus (40, 21.5%) followed by coagulase-negative staphylococci (CoNS; 31, 16.7%), P. aeruginosa (21, 11.3%) and E. coli (15 8%) (Table 3).

Table 3 The distribution of bacteria according to clinical diagnosis
Full size table

Risk factors for bacterial occurrence in ocular infections

In this study, 11 independent variables were considered during the bivariate analysis of risk factors for bacterial infection. In multivariate analysis, ocular surface disease (AOR = 13.6, 95% CI: 3.8 49.3, P < 0.01), ocular trauma (AOR = 4.2, 95% CI: 1.4–13, P = 0.012), hospitalization (AOR = 3.3, 95% CI: 1.03–12.5, P = 0.04), and cosmetic application practices (AOR = 4.7, 95% CI: 1.6–13.9 P < 0.01) were significantly associated with bacterial occurrence (Table 4). Though not statistically significant, bacterial occurrence was higher in housewives 32 (74.4%) (COR = 2.3, 95% CI: 0.6–9.0, P = 0.06) and in those subjects with no formal education 90 (73.8%) (COR = 1.8, 95% CI: 0.6–5.0, P = 0.2) (Table 5).

Table 4 Multivariate logistic regression analysis of factors associated with ocular bacterial infections
Full size table
Table 5 Bivariate logistic regression analysis of factors associated with ocular bacterial infections
Full size table

Antimicrobial susceptibility pattern of bacterial isolates

In this study, the majority of bacterial isolate were susceptible to ciprofloxacin (141, 89.2%), amikacin (137, 93.2%), gentamicin (131, 89.1%), chloramphenicol (106, 70.2%), and doxycycline (100, 71.9%). However, bacterial isolates were less susceptible to tetracycline (79, 51.3%) and SXT (66, 48.7%). Among the Gram-negative bacteria, 47 bacterial isolates (83%) were susceptible to ceftriaxone. In addition, 79 (70.7%) of the Gram-positive isolates were susceptible to erythromycin. All isolates of S. pyogenes, S. pneumoniae, viridans streptococci and H. influenzae were susceptible to clarithromycin. Methicillin-resistance was observed in 7 (17.5%) S. aureus and 14 (45.2%) CoNS isolates (Table 6).

Table 6 Antimicrobial susceptibility of bacteria isolated from ocular infections
Full size table

Overall, 40 (22.5%) bacterial isolates were resistant to at one antimicrobial agent; 96 (53.9%) were resistant to ≥2 antimicrobials (Table 7). A standard definition of MDR was applied for isolates of S. aureus, members of the Enterobacteriaceae and Acinetobacter spp. [43], by which 18 (45%) S. aureus, 17 (31.5%) isolates of Enterobacteriaceae, and one Acinetobacter isolate were found to be MDR. Among Enterobacteriaceae, Klebsiella spp. (6, 50%) and E. coli (7, 46.7%) were the commonest MDR isolates (Table 8).

Table 7 Level of antimicrobial resistance in bacteria isolated from ocular infections
Full size table
Table 8 Multidrug resistance patterns of bacterial isolates from ocular infections
Full size table

Discussion

Our assessment revealed that a high proportion of ocular infections (66.7%) were due to bacterial infections elsewhere in Ethiopia, lower proportions have been observed (48.8–60.8%) [10, 11, 44]; however, higher proportions (74.7%) have been observed in Jimma, Ethiopia [6]. This variation could be attributed to the inclusion of only exra-ocular infection in other studies. Our observations were also higher than that seen in India (34.5%) [3], Japan (32.2%) [4], and Iran (37.5%) [45]; sociodemographic, geographical or climatic differences for the patient populations could partially explain this [38]. Similar to findings from India [46], Iran [9], and other studies in Ethiopia [6, 10, 11, 44], Gram-positive bacteria contributed to majority (60.7%) of the total bacterial isolates in our study. Overall, the predominant bacterial isolate was S. aureus (21.5%), as has been reported in Jimma (28.4%) [6], Gondar (21%) [10], Nigeria (23.7%) [23], and India (26.6%) [47]. Staphylococcal isolates were predominant among patients with conjunctivitis, blepharitis and blepharoconjunctivitis diagnoses, similar to that in patients from Jimma [6], Nigeria [23], and Columbia [48]. Conversely,, CoNS isolates were more frequent in keratitis (17%) and endophthalmitis (28.6%) diagnoses, as has been reported in Uganda [49], Mexico [50], Australia [8], Pakistan [51] and India [52].

As compared to published literature from other areas in Ethiopia, Nigeria, UK, Australia and India, most streptococcal isolates were from conjunctivitis diagnoses, with some isolates from blepharoconjunctivitis and keratitis. In the literature, Staphylococci and streptococci are known to be the main agents of post-traumatic periorbital cellulitis [54]; in our study, S. aureus (50%), S. pneumoniae (12.5%) and S. pyogenes (12.5%) were the main Gram-positive isolates from periorbital cellulitis. As has been seen in Gondar [19], Borumeda [44], and India [47], S. pneumoniae and S. pyogenes were the predominant species among patients with dacryocystitis.

The most common isolates among the Gram-negative bacteria were P. aeruginosa (11.3%), E. coli (8.1%) and Klebsiella spp. (6.4%) as has been seen in Hawassa; however, the proportions of P. aeruginosa (4.9%) and E. coli (4.9%) observed in Hawassa were lower than in our study [23]. This may be due to the additional isolates from endophthalmitis diagnoses we identified.

P. aeruginosa was the predominant isolate in keratitis (23.4% of diagnoses) similar to findings reported in Nigeria (23.8%) [23], the UK (24.3%) [7], Australia (21%) [8], and Iran (24.7%) [15]. Pseudomonal keratitis is a progressive infection with large infiltrate and scarring [22, 53]; this therefore means majority of patients in our population are at risk of blindness.

Our assessment illustrated that the presence of ocular surface disease, ocular trauma, hospitalization and cosmetic application practices were significantly associated with the occurrence of bacterial infection. Risk factors for bacterial ocular infections that have been described include a history of hospital stay in Columbia and Portugal [48, 55], ocular trauma in Australia, Florida, Iran, China and Mexicao [1, 8, 15, 22, 50] and ocular disease in Florida and Colombia [22, 48] Cosmetic application practices were considered only for female participants. Limited information exists on the relationship between cosmetic application practices and bacterial ocular infections.

Almost one quarter of the bacterial isolates in were susceptible to all of the tested antimicrobials. Resistance to two or more antimicrobials was seen in 53.9% of the isolates, which is lower than that reported in other regions in Ethiopia (69.9–87.1%) [10, 11]. S. pneumoniae (87.5%), CoNS (87%) and S. aureus (82.5%) were the most resistant among Gram-positive isolates. Acinetobacter spp. (100%), Klebsiella spp. (100%) and Enterobacter spp. (80.0%) were the most resistant among Gram-negative isolates. which is lower than that reported in other studies [56, 57].

Here, relatively, amikacin (93.2%), gentamicin (89.1%) and ciprofloxacin (89.2%) were revealed high efficacy towards Gram-positive and Gram-negative isolates. Studies in Gondar have reported lower susceptibility to gentamicin (54.8%) and ciprofloxacin (74.2%) [10, 19]; differences in the variety of isolates tested may partly explain this. Outside Ethiopia, comparable susceptibility patterns to amikacin, ciprofloxacin and chloramphenicol were reported in Iran [15] and India [47]. However, the gentamicin susceptibility in this study is much higher than that in the study conducted in India (58.6%) [47]; regional variations in antibiotic prescription practices could explain this [38]. The resistance we observed to tetracycline and erythromycin was high, and this paralleled a study from India [3]. High resistance is usually due to the over use and empirical treatment of patients that ultimately leads to the emergence of drug resistant strains [58, 59].

Methicillin resistance was seen in 17.5% of the isolated S. aureus, which is in line with a study conducted in India (14%) [60], but lower than the rates in Taiwan (52.8%) [57] and Uganda (31.9%) [49]. In addition, 45.2% of CoNS were methicillin resistant, which was higher than detected in Uganda (27.6%) [49]. To date, other than a study conducted in Gondar, most studies in Ethiopia did not determine methicillin resistance rates for ocular staphylococcal isolates [19]. Our study was not without limitation; we did not isolate anaerobic bacteria or Chlamydia trachomatis due to unavailability of the prerequisite facilities. We also did not determine the antimicrobial susceptibility of Moraxella species or Aeromonas species due to a lack of antimicrobials for agar dilution.

Conclusions

In conclusion, a high prevalence of bacteria was found in patients with ocular infection. “Ocular surface disease, ocular trauma, hospitalisation and cosmetic application practices were significantly associated with the occurrence of bacterial infection. Overall, S. aureus, CoNS, and P. aeruginosa were the predominant isolates. The majority of the bacterial isolates were multidrug resistant. Methicillin resistance was higher in CoNS than in S. aureus. S. pneumoniae, CoNS of which S. aureus was the most resistant species among Gram-positive isolates; Klebsiella species and E. coli were most resistant among Gram-negative isolates. Identification of the specific etiologic agent and antimicrobial susceptibility testing should be practiced during the management of ocular infections to reduce the further emergence of multidrug-resistant bacteria”.