Common bacterial causes of external ocular infections, associated risk factors and antibiotic resistance among patients at ophthalmology unit of Felege Hiwot Referral Hospital, Northwest Ethiopia: a cross-sectional study

Background Ocular infections are a serious public health problem in Ethiopia with increased incidence of morbidity and blindness. Empirical therapy with topical ophthalmic broad spectrum antibiotic formulations is also a prevailing practice. The aim of this study was to determine the bacterial causes of external ocular infections (EOIs), antimicrobial resistance and its associated risk factors among patients at Felege Hiwot Referral Hospital, Northwest Ethiopia. Methods A hospital based cross - sectional study was conducted from 1 February to 30 April 2019. Patients with EOIs were consecutively included from 1 February to 30 April, 2019. Data were collected using structured questionnaire by face-to-face interview and complemented with patient card review. Conjunctival, eyelid margin and lacrimal sac swabs were collected aseptically. Bacterial species were identified using standard bacteriological techniques. Antimicrobial susceptibility testing was done using Kirby-Bauer disk diffusion method. Binary logistic regression analysis was calculated to identify the factors associated with EOIs. P.value (p) < < 0.05 was considered as statistically significant. Results A total of 360 patients took part in the study and majority of them were males (64.7%). The median age of study participants was 59.5 years. Overall, 208(57.8%) (95%CI = 52.6– - 62.8%) of patients had culture confirmed bacterial EOIs. The proportion of culture confirmed EOIs was 60.4% among conjunctivitis cases and 55.8% among blepharitis. Ocular trauma (P < < 0.001), ocular disease (P < < 0.001) and having eye allergy (P = 0.027) were significantly associated with EOIs. The most frequent isolates were S. aureus (37%), K. pneumoniae (13.5%), Proteus(7.2%) spp., S. pneumoniae (4.3%), Citrobacter spp. (3.4%) and P. aeruginosa (2.9%). Gram positive isolates were resistant to penicillin in 87% and ampicillin in 86.2%. Gram negative isolates were resistant to ampicillin (87.5%). P.aeruginosa isolates (85.3%) were resistant to piperacillin and 50% were resistant to tobramycin. Overall, 45.2% of the isolates were multi-drug resistant. The percentage of multi-drug resistance was 80% among Enterobacter isolates and 64.3% among K. pneumoniae. Conclusions Bacterial external ocular infections linked with multi-drug resistance and resistance to penicillin, ampicillin, tetracycline and piperacillin are high. Therefore, empirical treatment of eye infections in the study area needs to be monitored by regular antimicrobial-susceptibility testing of isolates.

spectrum antibiotics. However, widespread and misuse of these antibiotics for bacterial and viral infections or prophylactics has resulted in emerging global increase of antibiotic resistance [14].
Empirical treatments and self-medications with broadspectrum antibiotics which are responsible for selection of antibiotic resistance are routinely practiced for bacterial ocular infections in Ethiopia [11,15]. Thus, diagnosis of EOIs is without laboratory confirmation and little is known about specific etiologies in Ethiopia. As a result, patients are mostly managed empirically and multi-drug resistant bacteria isolates such as methicillin-resistant S.aureus are becoming very serious problem. Moreover, bacterial etiologic agents of ocular infections and its susceptibility profile to various antibiotics vary according to geography, regional location and from hospital to hospital. Therefore, this study identified the bacterial pathogens, determined their resistance profile to the commonly used antibiotics and identified factors associated with EOIs among patients in Ophthalmology unit of Felegehiwot Referral Hospital, Bahir Dar Ethiopia.

Study design, period and area
A cross-sectional study was conducted from February to April, 2019 at Felege Hiwot Referral Hospital, Ethiopia. Felege Hiwot Referral Hospital is one of the biggest tertiary level referral Hospitals in the Region visited by around 7 million peoples. The hospital currently delivers health care services in its different wards, intensive care units and Ophthalmology units with a total of 400 beds and 561 staffs [16]. On average, 100 patients attend daily at the ophthalmology unit for different ophthalmic cases to get secondary eye care, refraction examination, minor and major ocular surgery and other common examinations and treatment services. All patients with external ocular infections at Felege Hiwot Referral Hospital Ophthalmology unit during the study period were the study populations.

Inclusion and exclusion criteria
Patients who had redness of the eye, discharging, mucoid or mucopurulent secretion and/or conjunctival thickening were included to the study. But patients who received antibiotic treatment for the past 2 weeks and those who had case of keratitis were excluded from the study.

Variables
Bacteria profile of external ocular infections was the dependent variable while demographic variables (age, sex, and residence, educational and occupational status), types of ocular infections, previous ocular infection, eye surgery, ocular trauma, previous eye allergy and history of self-medication were independent variables.

Sample size determination and sampling
The sample size was determined using single population proportion formula n = (z) 2 P (1-p)/ d 2 where, n = sample size, z = level of confidence according to the standard normal distribution, p = sample proportion and d = tolerated margin of error. Therefore, by taking z = 1.96 for a level of confidence of 95%, p = 0.624 which is proportion of external ocular infection taken from previous study in Gondar, Ethiopia [15] and 5% margin of error. Thus, the sample size was calculated as n = (1.96) 2 × 0.624 (1-0.624)/ (0.05) 2 = 360. A total of 360 patients included in the study. Convenient sampling technique was used to include the study participants from the study population attending at Ophthalmology unit.

Data collection Demographic and clinical data
Data on demographic characteristics and history of ocular infection, previous ocular trauma, previous eye surgery, and history of self-medication were collected from each participant by a trained ophthalmic nurse with face-to-face interview and patient card review using a pretested structured questionnaire.

Ocular sample collection and transportation
The presence of external ocular infections were clinically assessed by ophthalmologist with thoroughly examination of all patients using a slit lamp microscope [8].
Conjunctival and eyelid swabs were collected using sterile cotton swab pre-moistened with sterile physiological saline [17]. Pus from lacrimal sac was collected using dry sterile cotton swab by applying pressure over the lacrimal sac and allowing the purulent material to reflux through the lacrimal punctum. In cases of acute lacrimal abscess or chronic dacryocystitis, pus was drain and taken on a dry sterile cotton swab [8]. All swabs were transferred into a tube that had 2 ml brain heart infusion broth (Oxoid, UK). All samples were labeled and transported to Microbiology Laboratory of Felege Hiwot Referral Hospital immediately. All ocular samples were collected by the ophthalmologist and ophthalmic nurse.

Culture and identification of bacterial isolates
The collected eye swab samples were inoculated from brain heart infusion broth to Blood Agar, Chocolate Agar and MacConkey Agar plates (Oxoid, UK) using sterile wire loops. All the agar plates were then incubated at 37°C for 24 h. Chocolate agar and Blood Agar plates were incubated within a candle-jar to facilitate CO2 tension. After 24 h of incubation, all plates were examined for bacterial pathogen growth. Identification of bacterial pathogens was made by colony morphology, Gram staining, enzymatic and biochemical tests. Catalase, coagulase, optochin (5 μg) and bacitracin (0.04 μg) tests were applied to identify and differentiate gram positive cocci, while biochemical tests, such as triple sugar iron agar, citrate utilization, oxidase test, lysine decarboxylase agar, urease and indole tests were used to identify gram negative bacterial pathogens [18].

Quality control
Structured questionnaires were used to collect the data. Supervision of the data collection was made regularly on daily basis and the collected data checked for completeness and accuracy. All ocular specimens were collected following standard operating procedures. The sterility of culture media was checked by incubating the media overnight before its use. The performances of all the prepared culture media and biochemical tests were checked by using American Type Culture Collection (ATCC) standard reference strains (E. coli ATCC25922, P. aeruginosa ATCC27853 and S. aureus ATCC 25923). The strains were used as a quality control throughout the study for culture, Gram staining and antimicrobial susceptibly testing.

Data analysis
Data were entered and analyzed using Statistical Package for Social Science 23 (IBM Corp Released 2011.IBM SPSS statistics. Armonk, NY: IBM Corp). Descriptive statistics were calculated to describe demographic, bacterial external ocular infections and their antibiotic resistance profiles. Bivariate analysis was done to see the association between dependent and independent variables. To determine independent predictors of bacterial external ocular infections, multivariable logistic regression analysis was done by taking variables whose p-value was < 0.21 in the binary logistic regression model. Crude Odds Ratio (COR) and Adjusted Odds Ratio (AOR) with 95% Confidence intervals (CI) were calculated. P-value of < 0.05 was considered statistically significant.

Ethical consideration
This proposal was reviewed and approved by College of Medicine and Health Science Institutional Review Board. Permission to conduct the study was obtained from FHRH. Detail information including the objective of the study was given to each study participants. Written informed consent was obtained from each study participants and for children assent was obtained from parents before they are asked to give data and sample. Confidentiality of the information kept. Positive findings were reported to ophthalmologists for treatment and any other care.

Participants' characteristics
A total of 360 patients with external ocular infections were enrolled in the study. Of them, 233 (64.7%) were males and 275 (76.4%) were from rural settings. The age range was 4-70 years (Median: 59.5). Two hundred sixty-four (73.3%) of the study groups were illiterate (Table 1).

Multivariable analysis of external ocular infections
On multivariable analysis, external ocular infections was significantly associated with previous ocular disease (AOR = 3.53, 95% CI = 2.175-5.731), eye allergy (AOR = 4.71, 95% CI = 1.191-18.59) and trauma (AOR = 9.97, 95% CI = 4.543-21.9). Patients who had history of ocular disease were 3.5 times more likely to have bacterial external ocular infections compared to the counters. Likewise, patients who had previous eye allergy were 4.7 times more likely to have bacterial external ocular infections compared to those who had no eye allergy. Patients who had trauma were also 10 times more likely to become culture positive for external ocular infections compared to their counter parts (Table 3).

Discussion
Bacterial causes of external ocular infections are a serious health problem and highly associated with resistance to antibiotics in developing countries [11,15]. It is becoming major public health concern in Ethiopia while 57.8% of patients had conjunctivitis, blepharitis and dacryocystitis in this study. This prevalence is comparable with studies in other parts of Ethiopia (47.7-59.4%) [2,4,7,20] and India (58.8%) [21]. However, higher percentages were found in Egypt (78.7%) [22] and China (82.78%) [1]. Conjunctivitis was the most common types of external ocular infections with blepharitis in the present study. This was consistent with reports from other parts of Ethiopia [7,10,23]. Moreover, blepharitis was the most common ocular infection with conjunctivitis in other areas of Ethiopia [3,15].
Indentifying the etiology and typical infectious organisms of dacryocystitis is important for the optimal management of patients [24]. In this study, 45.5% of clinically diagnosed dacrocyocystitis cases were caused by aerobic bacteria which are parallel with previous studies in Iran and Israel [20,24]. The remaining culture negative cases of dacryocystitis might be because of anaerobic and fastidious bacteria or fungal etiologic agents [20,24].
In this study, Gram positive cocci bacteria shared 66.3% of external ocular infections. This is parallel with other studies in Ethiopia (61.5% & 52%) [10,11], Egypt (58.9%) [22] and China (78.4%) [1]. The predominance of Gram positive cocci might be due to contamination of the injured eye from skin floras. In addition, S. aureus was the most frequent etiology of conjunctivitis, blepharitis, and blepharo-conjunctivitis in the present study. This is consistent with similar studies done in Ethiopia [11,23], Nigeria [25] and United Kingdom [26]. Anatomical disruption during cataract extraction and lens implantation might be a good opportunity for Staphylococci to elicit infection.
In this study, 27.3% and 18.2% of dacryocystitis was caused by Gram positive and Gram negative bacteria, respectively. The predominance of Gram positives over Gram negative bacteria could be due to more number of chronic than acute dacryocystitis cases. This finding is consistent with a study done in Iran where 54% of dacryocystitis were due to Gram positive bacteria while 24% were due to Gram negative bacteria [24]. However, in a similar study conducted at Israel, 61% of dacryocystitis was caused by Gram negative bacteria while 39% were due to Gram positive bacteria [20]. This showed that the etiologic agents of dacryocystitis vary from study area to study area and clinical presentations of dacryocystitis. The proportion of external ocular infections caused by Gram negative bacteria (33.3%) in the present study was consistent with other reports from Ethiopia (38.5%) [7] and India (35%) [27]. Among Gram negative bacteria isolates, K. pneumoniae was the leading isolate in the present study. This is concurrent with studies in Egypt [25] and Libya [28]. But in Saudi Arabia [5] and other parts of Ethiopia [11,15], P. aeruginosa was the most frequent isolate. In contrast, E. coli was reported as dominant bacteria from Gondar, Ethiopia [29].
The commonest organism isolated from cases of dacryocystitis in the present study was coagulase negative Staphylococcus that was isolated in 2 (18.2%) of positive cultures. S. aureus, Citrobacter spp., and P. aeruginosa accounted 9.1% each of the positive cultures. This is almost similar with a study done in Iran [24]. However, in a study done at Israel [20], Klebsiella was the commonest organism followed by E. coli, P. aeruginosa, Citrobacter and Enterobacter.
In this study, gram positive bacterial isolates revealed high percentage of resistance to ampicillin (86.3%) and penicillin (86.9%). This might be due to earlier exposure of the isolates to these drugs. These drugs are very common and patients can easily access them from pharmacies without a prescription. Similarly, 69.9-81.5% and 73.9-100% level of resistance to ampicillin and penicillin, respectively have been reported in other parts of Ethiopia [7,11,12].
S. aureus isolates were highly resistant to ampicillin (96.1%) and penicillin (96.1%). This finding is   [4]. The highest resistance percentage of S. aureus to penicillin and ampicillin might be due to production of beta-lactamase enzymes and alteration of the penicillin binding proteins [30]. The proportion of external ocular infections with methicillin-resistant S.aureus was 16.9% in the present study. This indicates the spreading of superbugs among outpatients in Ethiopia. The result is variable with previous reports from Uganda (31.9%) [31], United Kingdom (8.3%) [26], and United States (3-64%) [32].
In this study, majority (66.6%) of S. pneumoniae isolates were resistant to sulphamethoxazole-trimethoprim. This might be due to mutations in the dihydrofolate reductase and dihydropteroate synthetase genes [32]. Similar findings were reported in Hawassa, Ethiopia [16] and Nigeria [25] where 65% and 75% of S. pneumoniae isolates were resistant to sulphamethoxazole-trimethoprim, respectively.
All K. pneumoniae isolates of the present study were resistant to ampicillin. Proteus spp., Citrobacter spp., and Enterobacter spp., revealed 80-93.3% of resistance to ampicillin. This might be due to that these bacteria possess beta-lactamases that can confer resistance to ampicillin and many strains have acquired an extendedspectrum beta-lactamases [33]. The range of resistance was consistent with studies from other parts of Ethiopia where 80-100% of the above isolates were resistant to ampicillin [4,10,23].
All E. coli isolates of the present study were resistant to tetracycline. This might be due to multiple tetracycline efflux pumps and genetic exchange of resistance determinants among various clinical, environmental and commensal bacteria [34]. Similar findings were documented in other parts of Ethiopia [4,35]. Moreover, higher percentage of K. pneumoniae (60.7%) and Citrobacter spp. (66.7%) showed resistance to tetracycline in the present study.
Overall, the observed resistance of Gram positive and Gram negative species to different antibiotics could be linked with the empirical prescription of broad spectrum antibiotics, lack of regular screening of antimicrobial resistance, self-medications and misuse of antibiotics in the study area. These are the major contributors for the emergence and spread of multi-drug resistant isolates.
External ocular infections with multi-drug resistant isolates (45.2%) are a major concern in the present study which needs careful handling of eye wounds, trauma or surgery. The percentage is variable with earlier studies in Tigray (53.9%) and Gondar (66.4 and 87%), Ethiopia [9,12,15], Saudi Arabia (39%) [5] and China (12.1%) [1]. The percentage of multi-drug resistance isolates of S. aureus in the present study (45.5%) is lower than a study reported in Gondar, Ethiopia (64.6%) [23]. High percentage of multi-drug resistance was also documented in K. pneumoniae (64.3%) and Enterobacter spp. (80%) isolates. This could be associated with biofilm formation, transfer of resistance plasmid genes, efflux and production of extended-spectrum beta-lactamases which allows for resistance to multiple antimicrobials [36].
History of ocular trauma was a predictor variable for external ocular infections in this study. This is because mechanical disruption of the conjunctiva or stroma layers of eye allows the skin flora and other contaminants to breach, colonize and initiate infections. The finding was consistent with other studies in Ethiopia [4,9], Nepal [37], India [38] and Iran [39]. Moreover, previous ocular disease was another predictor variable for the occurrence of external ocular infections in the present S. pneumoniae (n = 9)  study. This finding was supported by other studies in Ethiopia [9], Nepal [37] and Iran [40]. Presence of eye allergy was strongly associated with bacterial external ocular infections in the present study. In most conjunctivitis infection, allergy is one of the major predisposing factors. Itching symptom of allergy might be a cause for skin contaminants to be introduced to the eye and cause chronic bacterial ocular infection [41].
This study was not without limitations because external ocular infections due to anaerobic bacteria, Chlamydia trachomatis and molecular mechanism of antimicrobial resistance were not addressed due to the limitations of laboratory setups. Moreover, cases of keratitis which requires special training and experience for corneal scraping were not included.

Conclusions
Conjunctivitis, blepharitis and dacryocystitis forms of bacterial external ocular infections linked with multidrug resistance and high levels of resistance to penicillin, ampicillin, tetracycline and piperacillin are prevalent in the study area. Therefore, empirical treatment of eye infections in the study area needs to be guided by antimicrobial -susceptibility testing. Previous ocular diseases, trauma and co-existence of eye allergy were predictor variables for bacterial external ocular infections. Bacterial isolates were susceptible for ciprofloxacin and gentamycin. Further studies on keratitis, intraocular infections and extents of beta-lactamase producing bacterial causes of external ocular infections using molecular techniques are required.