Pathogens and Antibiotic Susceptibilities of Global Bacterial Keratitis: A Meta-Analysis

Bacterial keratitis (BK) is the most common type of infectious keratitis. The spectrum of pathogenic bacteria and their susceptibility to antibiotics varied with the different regions. A meta-analysis was conducted to review the global culture rate, distribution, current trends, and drug susceptibility of isolates from BK over the past 20 years (2000–2020). Four databases were searched, and published date was limited between 2000 and 2020. Main key words were “bacterial keratitis”, “culture results” and “drug resistance”. Forty-two studies from twenty-one countries (35 cities) were included for meta-analysis. The overall positive culture rate was 47% (95%CI, 42–52%). Gram-positive cocci were the major type of bacteria (62%), followed by Gram-negative bacilli (30%), Gram-positive bacilli (5%), and Gram-negative cocci (5%). Staphylococcus spp. (41.4%), Pseudomonas spp. (17.0%), Streptococcus spp. (13.1%), Corynebacterium spp. (6.6%) and Moraxella spp. (4.1%) were the most common bacterial organism. The antibiotic resistance pattern analysis revealed that most Gram-positive cocci were susceptive to aminoglycoside (86%), followed by fluoroquinolone (81%) and cephalosporin (79%). Gram-negative bacilli were most sensitive to cephalosporin (96%) and fluoroquinolones (96%), followed by aminoglycoside (92%). In Gram-positive cocci, the susceptibility trends of fluoroquinolones were decreasing since 2010. Clinics should pay attention to the changing trends of pathogen distribution and their drug resistance pattern and should diagnose and choose sensitive antibiotics based on local data.


Introduction
Infectious keratitis (IK) is a potentially sight-threatening condition, which leads to at least 1.5 to 2 million new cases of unilateral blindness every year [1][2][3]. Among them, bacterial keratitis (BK) is the most common type according to the reports from multiple regions such as UK [4][5][6][7], North & South America [8][9][10][11], Middle East [12] and Australia [13,14]. Common risk factors for BK were contact lens wear, ocular trauma, ocular surface disease, and prior ocular surgery [15]. Bacterial culture via corneal scraping samples is still the gold standard for the diagnosis of BK, which permits isolation of the causal bacteria. However, not all medical institutions are able to carry out those tests due to various limitations. Therefore, it is critical for clinicians to make empirical diagnosis to know the pathogenic microorganism and antibiotics appropriate for eradicating the infection [16].
The common organisms that cause bacterial keratitis include Staphylococcus aureus, Coagulase-negative staphylococci (CoNS), Streptococcus pneumoniae, and others [17][18][19][20][21]. The bacterial spectrum and drug susceptibility for bacterial isolates from different areas or periods are widely reported [5,8,18,22,23], but the most common pathogen of BK remains debatable. Pseudomonas spp. were demonstrated to be the most common pathogen in Malaysia [22], Iran [23] and Taiwan [16], while CoNS are reported to be the most common in UK [5,6,24] and Australia [13]. Due to the widespread use of broad-spectrum antibiotics, it is very likely that the bacterial spectrum and its resistance to antibiotics varies greatly over time and from area to area. However, a comprehensive worldwide and long-term data analysis is scarce. To analyze the trends of bacteria and drug resistance profile over time in the world, we conducted a meta-analysis to compare the positive rate of culture in medical facilities worldwide and summarize the temporal and spatial trends of microbial isolates and their susceptibility patterns since 2000.

Literature Search and Study Characteristics
From the selected databases, 4734 potentially relevant references were identified. In total, 1156 references were excluded because of duplicates. Details of searching and de-duplications were shown in Appendix A. Search results were shown in Appendix B by reviewing the titles and abstracts, and 3459 references were excluded. After reading 119 full texts, 16 articles were excluded, and 103 papers were assessed for eligibility. A total of 42 studies were ultimately included in this meta-analysis, of which only 38 articles contained enough information for positive rate analysis, and 33 articles for drug susceptibility analysis. We used the methodological scoring system of "rate" to assess the quality of each study. The score of all studies were more than 4 points. The paper selection process was shown in Figure 1, and the characteristics of the included studies were shown in Table 1. We used the methodological scoring system of "rate" to assess the quality of each study. The score of all studies were more than 4 points. The paper selection process was shown in Figure 1, and the characteristics of the included studies was shown in Table 1. Among these data, 16 articles were reported from Asia, 11 from America, 1 from Africa, 9 from Europe and 5 from Oceania.
Antibiotics 2022, 11, 238 2 of 16 common in UK [5,6,26] and Australia [13]. Due to the widespread use of broad-spectrum antibiotics, it is very likely that the bacterial spectrum and its resistance to antibiotics varies greatly over time and from area to area. However, a comprehensive worldwide and long-term data analysis is scarce. To analyze the trends of bacteria and drug resistance profile over time in the world, we conducted a meta-analysis to compare the positive rate of culture in medical facilities worldwide and summarize the temporal and spatial trends of microbial isolates and their susceptibility patterns since 2000.

Literature Search and Study Characteristics
From the selected databases, 4734 potentially relevant references were identified. In total, 1,156 references were excluded because of duplicates. Details of searching and deduplications were shown in Appendix A. Search results were shown in Appendix B by reviewing the titles and abstracts, and 3459 references were excluded. After reading 119 full texts, 16 articles were excluded, and 103 papers were assessed for eligibility. A total of 42 studies were ultimately included in this meta-analysis, of which only 38 articles contained enough information for positive rate analysis, and 33 articles for drug susceptibility analysis. We used the methodological scoring system of "rate" to assess the quality of each study. The score of all studies were more than 4 points. The paper selection process was shown in Figure 1, and the characteristics of the included studies were shown in Table 1. We used the methodological scoring system of "rate" to assess the quality of each study. The score of all studies were more than 4 points. The paper selection process was shown in Figure 1, and the characteristics of the included studies was shown in Table 1. Among these data, 16 articles were reported from Asia, 11 from America, 1 from Africa, 9 from Europe and 5 from Oceania.

Positive Rate of Culture
A total of 38,931 samples scraping from the cornea of BK patients were reviewed in the study. Among them, 14,596 samples were culture positive and positive rate of culture was 47% (95%CI, 42-52%) based on the 38 studies ( Figure 2). The highest positive rate was 83% and the lowest positive rate was 21%. Data of positive rate was available among 21 countries (35 cities) in 5 continents. The highest and the lowest positive rate analyzed by counties were from Korea and from Turkey (83% vs. 28%, Figure 3, Table S1). There were no significant differences of positive culture rate among different countries (p = 0.464). Grouped by continents, the highest positive rate was 59% (95%CI, 48-68%), found in 4 articles performed in Oceania and the lowest was 40% (95%CI, 32-49%), found in 14 articles performed in Asia. However, there were no significant differences of positive culture rate between continents (p = 0.211).

Distribution of Bacteria Isolated from Corneal Lesions
Within 14,596 samples reported from 38 studies, 15,350 bacterial strains isolate corneas of BK cases were summarized (Table 2). Gram-positive cocci were the maj

Antibiotic Susceptibility of the Bacterial Strains Isolated from Corneal Lesions
All results of drug susceptibility tests of the strains were summarized in Figure  Most Gram-positive cocci were susceptible to aminoglycoside (86%, 3916 sensitive 4527), following cephalosporin and fluoroquinolone, 79% (1997 sensitive in 2515) and 81 (3921 sensitive in 4831), separately. A resistance of macrolides was observed (57%, 2

Antibiotic Susceptibility of the Bacterial Strains Isolated from Corneal Lesions
All results of drug susceptibility tests of the strains were summarized in Figure 5. Most Gram-positive cocci were susceptible to aminoglycoside (86%, 3916 sensitive in 4527), following cephalosporin and fluoroquinolone, 79% (1997 sensitive in 2515) and 81% (3921 sensitive in 4831), separately. A resistance of macrolides was observed (57%, 212 sensitive in 375). As for Gram-negative bacilli, most isolates were susceptible to cephalosporin and fluoroquinolone (cephalosporin: 96%, 1269 sensitive in 1328; fluoroquinolone: 96%, 2519 sensitive in 2611), followed by aminoglycoside (92%, 2547 sensitive in 2783). Figure 6 shown the changing trends of susceptibility of pathogens to common drugs. In Grampositive cocci, the susceptibility to common antibiotics such as cefazolin, gatifloxacin, moxifloxacin and ofloxacin showed a decreasing trend since 2010. For Gram-negative bacilli, a susceptibility over 90% was maintained in all recommended antibiotics [24] in recent years. sporin and fluoroquinolone (cephalosporin: 96%, 1269 sensitive in 1328; fluoroquinolone: 96%, 2519 sensitive in 2611), followed by aminoglycoside (92%, 2547 sensitive in 2783). Figure 6 shown the changing trends of susceptibility of pathogens to common drugs. In Gram-positive cocci, the susceptibility to common antibiotics such as cefazolin, gatifloxacin, moxifloxacin and ofloxacin showed a decreasing trend since 2010. For Gram-negative bacilli, a susceptibility over 90% was maintained in all recommended antibiotics [27] in recent years.

Discussion
Bacterial keratitis is the second most common cause of legal blindness worldwide [28]. Although broad-spectrum antibiotics, such as levofloxacin, have always been used to control BK, more targeted treatment was required to improve the clinical outcomes [29]. However, the spectrum of pathogenic bacteria and their susceptibility to antibiotics varied with the different regions. Kaye et al. showed that these variations were related to the latitude and the degree of urbanization of the population studied [29]. Therefore, local epidemiology of bacterial spectrum and its resistance to antibiotics should be paid attention and is mandatory to know. In this study, 30-year changing trends of microbiological profile of BK were reviewed.
From our results, the average rate of positive culture from the samples of BK was 47%. The large difference in positive culture rates (21~83%) between literatures may be related to the different indications of corneal scrapes in different medical institutions, the ability of microbiology laboratory, or to the fact that the study population had already

Discussion
Bacterial keratitis is the second most common cause of legal blindness worldwide [25]. Although broad-spectrum antibiotics, such as levofloxacin, have always been used to control BK, more targeted treatment was required to improve the clinical outcomes [26]. However, the spectrum of pathogenic bacteria and their susceptibility to antibiotics varied with the different regions. Kaye et al. showed that these variations were related to the latitude and the degree of urbanization of the population studied [26]. Therefore, local epidemiology of bacterial spectrum and its resistance to antibiotics should be paid attention and is mandatory to know. In this study, 30-year changing trends of microbiological profile of BK were reviewed. From our results, the average rate of positive culture from the samples of BK was 47%. The large difference in positive culture rates (21~83%) between literatures may be related to the different indications of corneal scrapes in different medical institutions, the ability of microbiology laboratory, or to the fact that the study population had already received topical antibiotics treatment before corneal scraping.
In this study, the common bacterial isolates were Staphylococcus spp., Pseudomonas spp., Streptococcus spp., Corynebacterium spp. and Moraxella spp., which was consistent with studies in the USA [27], UK and Canada. Although the number of Moraxella strains was 311, lower than that in Serratia strains (373), Moraxella spp. were isolated from 7121 samples (4.4%, 16 studies) and 373 Serratia spp. were isolated from 10,431 samples (3.6%, 27 studies). Thus, we concluded that Moraxella spp. were more common than Serratia spp. Pseudomonas spp. was demonstrated to be the most common pathogen in Singapore and Malaysia, which may result from a local high frequency of using contact lenses. Keya et al. found the percentage of Enterobacterales was 15.3%, higher than the percentage of Staphylococcus aureus and Pseudomonas aeruginosa. In addition, the upward trend of Gram-positive cocci and the downward trend of Gram-negative bacilli were observed, especially for the upward trend of Staphylococcus spp. and the downward trend of Pseudomonas spp. at the genus level. Similar trends were also presented in the studies from the UK [6] and Iran [23]. CoNS, one of the most common strains of Staphylococcus spp., was the major bacteria of normal skin, including eye lid. It would have more opportunity to contaminate the cornea; in addition, the conduction of corneal scraping was also easily susceptible to its contamination. The decreasing percentage of Pseudomonas spp. may be attributed to the wide application of some antibiotics, such as tobramycin and fluoroquinolones, and to the improvement in health conditions. Though several studies [4][5][6] reported a significant increase of Moraxella keratitis in UK for the last two decades, the trend was not shown in this worldwide study. Perhaps this trend was limited to specific regions.
In Gram-positive cocci, most isolates were susceptible to aminoglycoside (86%, 3916 sensitive in 4527), and resistant to macrolides in more than half of drug susceptivity tests (57%, 212 sensitive in 375). In the other two classes of antibiotics commonly used in clinical practice against Gram-positive cocci, cephalosporin and fluoroquinolone, 79% (1997 sensitive in 2515) and 81% (3921 sensitive in 4831) isolates showed susceptivity. Since 2010, the susceptibility of Gram-positive cocci to common antibiotics such as cefazolin, gatifloxacin, moxifloxacin and ofloxacin has shown a decreasing trend. For Gram-negative bacilli, most isolates were susceptive to cephalosporin (96%, 1269/1328) and fluoroquinolones (96%, 2519/2611), followed by aminoglycoside (92%, 2547/2783). The trends of susceptibility seemed stable and maintained above 90%. Some studies had reported a group of Pseudomonas spp. with multiple drug resistance [28,29]. It is necessary to use targeted antibiotics in case of the development of resistant strains.
The meta-analysis revealed the trend and distribution of bacterial keratitis pathogens and their drug resistance pattern guiding ophthalmologists to diagnosis and to choosing antibiotics based on their local data. Our study also possessed some limitations. Original studies often reported their results via time period (e.g., 2000-2005 CoNS 30%), not specific year (e.g., 2000 CoNS 30%), which affected the accuracy of our study. Differences between original studies, such as culture methods, participants who received topical antibiotics treatment before corneal scraping and experience of clinics would increase the difficulty of pathogen distribution analysis. Our study could still provide useful guidance for clinics.
In conclusion, the worldwide average positive culture was 47% between 2000 to 2020. The percentage of Gram-positive cocci was increasing, and the percentage of Gram-negative bacilli was decreasing. The five most common bacterial organism were Staphylococcus spp., Pseudomonas spp., Streptococcus spp., Corynebacterium spp. and Moraxella spp. Increasing trends of Gram-positive cocci and the decreasing trends of Gram-negative bacilli were observed in 1996-2015. Most Gram-positive cocci were susceptive to aminoglycoside and were resistant to macrolides. Gram-negative bacilli were sensitive to cephalosporin, fluoroquinolone and aminoglycoside and maintained susceptibility above 90%. The decreasing trends of susceptibility were observed in Gram-positive cocci to most common antibiotics. Ophthalmologists should pay attention to the changing trends of pathogen distribution and their drug resistance patterns and modify the diagnosis and choose sensitive antibiotics based on the local data.

Databases and Search Strategy
Four databases, including Embase, Medline, Web of Science, and CINAHL, were searched, and publication date of articles were limited between January 2000 to December 2020. Main key words were "bacterial keratitis", "culture results" and "drug resistance". The whole search strategy was ("bacterial keratitis" OR "infectious keratitis" OR "microbial keratitis" OR "bacterial infections" OR "corneal ulcers" OR "bacterial infections of cornea") AND ("organisms" OR "culture results" OR "isolates" OR "microbiology" OR "antibiotic susceptibility" OR "resistance pattern"). In addition, the document type was restricted to "article", the language was restricted to "English", and the subjects were restricted to "human". More details of the search strategy could be found in Supplement Materials Table S1.

Literature Selection and Quality Assessment
The literature searched in the above databases was imported into the EndNote X 9 software library for merging and de-duplicating; then, the titles and abstracts were screened by two researchers (Z.Z., J.L.) according to the following inclusion criteria: (1) Purpose of the article should concentrate on reporting the distribution and resistance pattern of bacterial keratitis isolates; (2) Subjects of the literature must contain patients suspected of infective keratitis and confirmed by positive bacterial culture results; (3) Pathogens were isolating from corneal scrapes via culture and were identified at least at the genus level; (4) Drug susceptibility tests should be conducted for isolated strains via in vitro minimum inhibitory concentration testing (MIC) or the disk diffusion method. The homogeneity of our meta-analysis was controlled by exclusive criteria below: (1) Subjects with small sample size or specific risk factors; (2) Subjects already selected by authors would be excluded for positive rate analysis. (3) Multiple articles published using the same data would be deduplicated. After preliminary exclusion of unrelated references, we downloaded the full text of each citation; the literature quality was evaluated based on a methodological scoring system of "rate" [30]. The detailed quality criteria were as follows: (1) Whether there is a clear diagnostic basis for bacterial keratitis; (2) Whether the sample size (the number of bacterial culture specimens) is greater than 246 cases; (3) Whether there are clear criteria for positive bacterial culture; (4) Whether there are clear study parameters, such as positive rate or drug susceptibility results; (5) Whether the data is complete. Complete data should include the description of study populations, methods for the drug susceptibility test, and protocol for bacteria separating and identifying. Each criterion was given one score, and studies with a score of 4 or more were of high quality and included for analysis. All the steps of screening and quality assessment were carried out independently by two researchers (Z.W., X.X.). In case of disagreement over the inclusion of the literature, a third, more experienced researcher (Q.L.) would make the final decision.

Data Extraction
According to the purpose of this study, the data extraction scale of the literature was developed. For each included article, four aspects of information would be extracted. The first is the basic information of the literature and the institution of the author, including the publication year, time that the research was conducted, author, title, medical institution, etc. Next, the necessary data of positive rate of culture, bacterial strains distribution and drug susceptibility were extracted. The positive rate of culture was defined as the proportion of BK patients with positive culture results in all culture-treated patients suspected of infective keratitis. The parameter necessary for analyzing bacterial strain distribution contained the total number of strains isolated, number classified by Gram staining and number of strains as accurate as possible to species. Parameters related to drug susceptibility included the number of susceptible or resistant species performed on a certain species of bacterium to a certain drug. All above data were extracted into Microsoft Excel software.

Statistical Analysis
Data were analyzed with SPSS software (SPSS for windows, version 16.0, SPSS, Chicago, IL, USA). The meta-analysis was conducted using R program (V.4.0, R Foundation for Statistical Computing, Vienna, Austria) with meta package. All effect sizes were transformed into a single common metric, event rate with its 95% confidence interval, which indicated the number of participants in each sample endorsing bacterial keratitis. Either a random effects model or a fixed effects model was used to perform meta-analysis, which was determined by I 2 statistic; I 2 > 50% indicates a large heterogeneity among included studies and, correspondingly, a random effects model would be used; otherwise, a fixed effects model would be used.