Article Text

Geographic variations in microbial keratitis: an analysis of the peer-reviewed literature
1. Ameet Shah1,
2. Arun Sachdev2,
3. David Coggon3,
4. Parwez Hossain4
1. 1Moorfields Eye Hospital, London, UK
2. 2St Paul's Eye Unit, Royal Liverpool University Hospital, Liverpool, UK
3. 3MRC Epidemiology Resource Centre, University of Southampton, Southampton, UK
4. 4Division of Infection, Inflammation & Immunity, University of Southampton, Southampton General Hospital, Southampton, UK
1. Correspondence to Mr Parwez Hossain, Division of Infection, Inflammation & Immunity, Eye Unit, University of Southampton, Southampton General Hospital, Tremona Road, Southampton SO16 6YD, UK; parwez{at}soton.ac.uk

## Introduction

Microbial keratitis is a potentially serious corneal infection and a major cause of visual impairment worldwide. A conservative estimate of the number of corneal ulcers occurring annually in the developing world alone is 1.5–2 million.1 Permanent visual dysfunction has been reported in a significant proportion of patients in both developing2 and developed3 countries. Srinivasan et al4 comment that ulceration of the cornea in south India ‘is a blinding disease of epidemic proportions.’

Various micro-organisms can cause microbial keratitis, and predisposing risk factors vary from one geographic region to another. They include pre-existing corneal disease as well as other risk factors such as contact-lens wear, surgical or non-surgical trauma, and ocular surface disease.5–7

There is limited comparative information on international patterns of causative organisms in microbial keratitis. With increasing rates of migration and international travel, an awareness of these geographical variations is relevant for clinicians treating microbial keratitis, and especially for those planning to work in regions where they have not previously practised. The aim of this review is to summarise the published literature that provides information on the worldwide variation in organisms causing microbial keratitis.

## Methods

### Search strategy

A systematic review of the current literature pertaining to the prevalence of causative organisms responsible for microbial keratitis was conducted. PubMed searches were performed and verified in April 2009 by two independent investigators. The terms ‘microbial keratitis,’ ‘bacterial keratitis’ and ‘infectious keratitis’ were entered into PubMed. Only papers presenting data that were collected after 1 January 1990 were examined, and the search was restricted to English language and human studies. Only studies that cultured at least 50 organisms in total were included. Titles and abstracts were read, and a judgement was made as to whether the paper provided culture results for microbial keratitis in a specified geographical location. If this was felt to be the case, a full text request was made to access the original published data.

Studies looking only at infections related to use of contact lenses were excluded, as were studies looking only at limited age groups.

### Extraction and recording of data

Papers were read, and information was abstracted on the following variables: number of patients in the study, time period of reporting, region, method by which organisms were isolated, method of culture, rate of positive cultures and number of contact-lens wearers in the study sample. These data were then entered into a Microsoft Excel spreadsheet.

With regard to the micro-organisms cultured, the total numbers in each of the following categories were abstracted and recorded: Gram-positive organisms—staphylococcal species and streptococcal species; Gram-negative organisms—Pseudomonas species; protozoa; fungi/yeasts—Aspergillus species and Candida species.

### Classification of income levels and GNI subheading

The prevalence of different causative organisms was compared according to countries' gross national incomes (GNIs) (source=http://web.worldbank.org). Income groups were defined by 2007 GNI per capita, calculated using the World Bank Atlas method.8 The groups distinguished were: low income, $935 or less; lower middle income,$936–$3705; upper middle income,$3706–$11 455; and high income,$11 456 or more.

### Statistical analysis

Statistical analysis was performed using ‘Analyse-it’ version 2.20 software. Spearman correlation coefficients were used to explore associations:

1. between prevalence of certain types of organism and GNI; and

2. between prevalence of contact-lens wear and prevalence of pseudomonas.

## Results

A total of 3883 publications were identified through the preliminary PubMed search. Of these, 37 papers met the inclusion criteria. One paper was excluded because it included a significant number of cases that the authors deemed to constitute an outbreak of suture-related infections.9 Twelve of the included papers were from the Indian subcontinent, seven from North America and Canada, six from the Far East, five from Australasia, four from Europe, two from Africa (both from Ghana) and one from South America. The median GNI of the countries studied was $22 685 (range$470–$59 880). The number of patients ranged from 73 to 3183. The time periods of study ranged from 3 to 192 months, although three studies did not specify the study period. The proportion of keratitis patients with a recent history of contact-lens wear was reported in only 22 studies and ranged from 0.33% (West Bengal10) to 50.3% (Paris11). Three studies reported on only culture-positive cases and so appear to have 100% culture-positive rates in table 1. In the remainder of the studies, culture-positive rates ranged from 35% to 86%. Table 1 Studies meeting criteria for inclusion in review Bacterial isolates from each study are summarised in table 2. Among studies which looked at non-bacterial as well as bacterial organisms, Los Angeles33 and Adelaide29 had the highest percentages of bacterial cases (95% in both), with Paraguay22 having the highest percentage of staphylococcal species (79%), and Bangkok27 the highest proportion of pseudomonal infections (55%). Tamil Nadu12 had the highest percentage of streptococcal infections (47%). The highest percentage of protozoal infections (7%) was found in a study from Hong Kong.5 Table 2 Bacterial isolates by location Protozoal and fungal/yeast isolates from each study are summarised in table 3. Eastern India10 had the highest proportion of corneal infections attributable to fungi (67%). When considering those countries with a significant proportion of fungal ulcers (we have arbitrarily chosen a cut-off of 10% or more), East India also had the highest percentage of aspergillus (60% of all fungal cultures), whereas the highest percentage of fusarium (73% of all fungal cultures) was found in a study from Hyderabad.18 Table 3 Protozoa and fungal/yeast isolates by location Statistically significant correlations were found between GNI and percentages of bacterial, fungal (see figures 1, 2) and streptococcal isolates (data not shown). The Spearman correlation coefficient for prevalence of bacteria and GNI was 0.85 (95% CI 0.68 to 0.91, p<0.0001), for prevalence of fungi and GNI −0.81 (95% CI −0.90 to −0.66, p<0.0001) and for prevalence of streptococci and GNI −0.43 (95% CI −0.66 to −0.12, p=0.009). Surprisingly, there was no statistically significant correlation between percentage of pseudomonal isolates and percentage of contact-lens wearers. Figure 1 Scatter plot showing the percentage of bacterial isolates in studies not looking exclusively at bacterial causes of microbial keratitis plotted against gross national income (US$ per capita).

Figure 2

Scatter plot showing the percentage of fungal isolates plotted against gross national income (US\$ per capita).

## Discussion

We have found a wide variation in the causative organisms for microbial keratitis in different parts of the world. To some degree, this variation is explained by economic factors as well as contact-lens wear. A high proportion of bacterial ulcers were reported from centres in developed countries (North America, Australia and Western Europe). In these countries, patients are far less likely to be agricultural workers, and so have a reduced risk of trauma from organic matter, which is known to be a risk factor for fungal infection.25

A high percentage of staphylococcus species (79%) was recorded in the study from Paraguay,22 although the reason for this is not clear. Of note, the authors comment that their patients have to make long journeys to their hospital. Thus, their data may reflect more severe cases of microbial keratitis.

The study from Tamil Nadu12 found the highest proportion of streptococcus species (46.8%). The authors noted that this figure was only 18.5% in 1986 and suggest that the trend might represent a genuine change in the bacterial flora owing to changes in the climate and environment.

The study from Bangkok27 had the highest proportion of pseudomonas infections (55%). Interestingly, this study did not have the highest proportion of contact-lens wearers (only 24%). Other studies reported far higher proportions of contact-lens wearers—for example, 44% in a study from Taiwan23 and 50% in the study from Paris.11 When we compared the percentage of contact-lens wearers with the percentage of pseudomonal infections, the Spearman correlation coefficient was not statistically significant. Interestingly, Cohen et al39 at Wills Eye Hospital reported a decline in contact lens-related ulcers: during 1998 to 1991, contact-lens wear accounted for 44% of all ulcers, but during 1992 to 1995, it accounted for only 30%. The authors speculated that their figures might reflect a reduction in the number of referrals to their unit due to the increased availability of fluoroquinolones in the community.

Trauma was a major risk factor for corneal infection in certain countries. In Paraguay,22 the percentage of cases with preceding trauma was 48%, in Eastern Nepal,14 53%, in Madurai, South India,4 and 65% and 83% in Eastern India10 (most commonly from injury by the paddy or its stalk). The authors of this last study noted an increase in keratitis during the harvesting season.

The above studies also addressed the frequency of self-medication prior to presentation at a tertiary referral unit. In the Madurai study, 20% of patients had been to a village healer, and 87% had been started on topical medication, of whom 8% were on topical corticosteroids. In the study from Eastern India, 18% of patients had used medication before coming to the clinic, and in the Paraguay study, the proportion was 83%. Schwab40 states that in some developing countries, locally produced over-the-counter antibiotics may be cheaper than imported preparations, and this may have an impact on the prompt use of appropriate therapy. Access to these over-the-counter antibiotics is limited, and individuals may be more inclined to go to local healers who are generally well respected but may instil harmful ‘traditional’ agents such as vegetable extracts, leaves, herbs, human urine or animal products, or perform practices such as thermal cautery.

Jeng and McLeod41 commented on the emerging resistance of bacterial infections to fluoroquinolones. In addition to changes in resistance patterns, studies have also demonstrated changing patterns of causative organisms over time in a given geographical location. Varaprasathan et al42 reported that the proportion of Streptococcus pneumoniae and Pesudomonas aeruginosa ulcers in Northern California had decreased over a 50-year period, while that of Serratia marcescens had increased over the same period. Sun et al43 reported a rise in the percentage of Gram-positive cocci in North China from 25% in 1991 to 70.8% in 1997, as well as a decrease in Gram-negative bacilli from 69% to 23.4% over a similar period.

Leck et al12 have previously compared corneal ulcers in Ghana and South India, while Lam et al5 have discussed differences between Hong Kong, Europe and North America. However, the present study is the first to present a worldwide comparison of corneal infections.

In interpreting this comparison, a number of limitations must be considered. Variations existed in the definition of microbial keratitis between studies. Lam et al, reporting on cases from Hong Kong,5 included patients with ‘the clinical presentation of a corneal stromal infiltrate >1 mm2.’ This differs from Srinivasan et al,4 who included patients with ‘loss of the corneal epithelium with underlying stromal infiltration and suppuration associated with signs of inflammation with or without hypopyon.’ There were variations in methods of culture. For example, one study16 used sheep's blood agar, chocolate, non-nutrient, Sabauroud, brain–heart infusion and potato dextrose agar, while another13 used only chocolate and Sabouraud media. Some studies did not specify the media used.17–19 All studies included bacterial infections, but not all included fungal, protozoal and yeast organisms. The majority of studies looked at all cases of microbial keratitis, while some looked only at patients requiring hospital admission (Wong et al and Cheung et al3 30). It is likely that in these studies, particularly virulent organisms will be over-represented. Finally, data are only available from centres that have conducted studies on microbial keratitis, limiting the coverage of certain regions of the world.

Despite these limitations, we have presented to our knowledge, for the first time, a worldwide overview of causative organisms in microbial keratitis demonstrating associations between specific types of microbial keratitis and national income.

View Abstract

## Footnotes

• Competing interests None.

• Provenance and peer review Not commissioned; externally peer reviewed.

## Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.