Causes of suppurative keratitis in Ghana

1024 British J7ournal of Ophthalmology 1995; 79: 1024-1028 Causes of suppurative keratitis in Ghana Br J Ophthalmol: first published as 10.1136/bjo.79.11.1024 on 1 November 1995. Downloaded from http://bjo.bmj.com/ on June 14, 2020 by guest. Protected by copyright. Maria Hagan, Elizabeth Wright, Mercy Newman, Paul Dolin, Gordon Johnson Abstract 30% are fungi,5 and in Nepal 17%.6 In Aims-Suppurative keratitis is a serious temperate climates such as Britain7 and problem in all tropical countries, but very northern United States,8 the proportion of little information is available about the fungi causing suppurative keratitis is very causative organisms in Africa. The objec- small. Similarly, at the high altitude of tives were to identify the causative organ- Johannesburg, South Africa, between 2 1% isms and the proportion of cases caused and 2-3%9 10 were caused by fungi, represent- by fungi in southern Ghana, and to deter- ing very few individual cases. There is a report mine whether correct decisions about of 21 cases of mycotic keratitis from Nigeria in treatment could be made on the basis of 1976,"1 but very little information is otherwise Gram stain in the eye clinic. available from sub-Saharan Africa. Methods-Scrapings were taken from Usually no appropriate antibiotics or corneal ulcers of consecutive new patients ophthalmic antifungal agents are available, presenting at Korle Bu Hospital, Accra, especially for treating an ulcer at a district and inoculated on 'chocolate' and level. Yet, paradoxically, ophthalmic cortico- Sabouraud's agars. Further scrapings steroid preparations are freely available in were taken for Gram staining and interpre- some countries. An additional factor is that tation in the eye clinic. Duplicate slides many people go to the traditional healer first, were assessed by an experienced micro- resulting in further delay and sometimes dam- biologist in the UK. age to the cornea. Results-One or more organisms were A useful development has been the demon- cultured from 114 of 199 patients (57.3%), stration by Williams and associates that a simple the most common being Fusarium microbiological laboratory could be established species, Pseudomonas aeruginosa, and in Bangladesh and make a substantial difference Staphylococcus epidermidis. Fungi, alone to accuracy of management of corneal suppura- or in combination, were isolated from 56% tion.12 Of 58 cases which were culture positive of the patients who had positive cultures. the results could have been anticipated in 47 on In total, 122 patients (61.3%) had their the basis of Gram stain alone. treatment either determined or altered In Ghana blinding suppurative keratitis is a based on the results of the microbiological major problem. At the same time approximately diagnosis; in 87 of these solely on the basis 120 nurses have now completed a 1 year train- of direct microscopic examination. ing for an ophthalmic nursing diploma, offering Conclusions-Infection by filamentous potential for greatly improved primary and fungi accounted for more than half of the secondary care. These graduates are working ulcers from which cultures were obtained. throughout the country in an extended role, Both training in technique and experience often providing the only secondary eye care in in interpretation are necessary for district hospitals. It is, therefore, important that microscopy based diagnosis by staff in the they are assisted to have a logical approach to clinic to be of greatest value. Direct these ulcers and to be provided with some microscopy was particularly useful for essential medication for prophylaxis and treat- detecting fungi. ment. Korle Bu Teaching (Br_J Ophthalmol 1995; 79: 1024-1028) The primary purpose of this investigation Hospital, University of has, therefore, been to determine the actual Ghana, Accra, Ghana organisms causing suppurative corneal ulcera- M Hagan M Newman Suppurative keratitis (suppurative corneal tion, and the proportion of cases caused by ulceration) is a serious problem in most tropi- fungi, in southern Ghana. The second objective Worthing Hospital, cal countries.1 In population based surveys in was to find out for how many cases a correct Park Avenue, Worthing, England Africa, corneal opacification has usually been decision regarding treatment could be made in E Wright the second cause of blindness after unoperated the eye clinic on the basis of Gram stain alone. Institute of cataract. Many of these cases represent the The long term aim is to develop and Ophthalmology, long term sequelae of trachoma, but often test appropriate and practical methods of Department of suppurative infection is superimposed on prophylaxis and management for suppurative Preventive damage due to trachoma. A proportion of corneal ulceration at a community level, using Ophthalmology, cases recorded as phthisis are the result of the London paramedical and primary health care workers P Dolin G Johnson perforation of infected ulcers, and others of the and a simple protocol with a limited choice of infection of injuries. Filamentous fungi are medications. Correspondence to: Dr Gordon Johnson, responsible for a larger proportion of these Institute of Ophthalmology, corneal infections in tropical latitudes than in Department of Preventive temperate climates. In south Florida fungi Methods Ophthalmology, Bath Street, account for 35% of the isolates in microbial London EC1V 9EL. Accepted for publication keratitis.2 In Bangladesh the proportion is SUBJECTS 9 August 1995 between 36% and 40%,3 4 in southern India Consecutive new patients presenting to the Causes of suppurative keratitis in Ghana 1025 Eye Unit of Korle Bu Hospital, Accra, were CLINICAL MANAGEMENT entered into the study if they had clinical signs A fixed protocol was established for initial of established suppurative corneal infection treatment with both topical medication and Br J Ophthalmol: first published as 10.1136/bjo.79.11.1024 on 1 November 1995. Downloaded from http://bjo.bmj.com/ on June 14, 2020 by guest. Protected by copyright. with loss of epithelium over at least 2 mm subconjunctival injection based on the Gram diameter and underlying stromal infiltration. stain result. Patients were excluded if they refused investi- Rules were also established for initial gation and treatment, had viral ulcers which treatment if no organisms were seen on the were not secondarily infected, Mooren's ulcer smear or the identification of the bacteria was or other peripheral ulcers, had had recent per- in doubt. There were similar guidelines for forating trauma, were already under treatment modifications to the initial treatment, based on at the department, or were neonates less than culture and sensitivity results or failure of 28 days post partum. response to initial treatment. CLNICAL EXAMINATION FURTHER IDENTIFICATION OF ORGANISMS When accepted into the study, each patient The culture plates and bottles were taken imme- was assigned a number and a separate record diately to the microbiology laboratory at Korle Bu form in addition to the regular hospital Hospital for incubation, identification, and test- records. The patient's age, sex, occupation, ing ofsensitivities according to standard methods. and place of residence were entered. A history The duplicate slides were mailed in batches to the was taken of the circumstances in which the microbiologist in the UK. Initial cultures eye became infected, of predisposing factors, where fungi were suspected and secondary cul- and any prior treatment received. tures of fungi grown in the laboratory and inoc- Using a slit-lamp, a qualified ophthal- ulated on Sabouraud slopes, were mailed, or mologist or ophthalmic medical officer taken in batches to the mycologist in London. examined each case and made a drawing on Where an eye was irretrievable, cultures the record form of the appearance when first were taken and the comeal disc was removed seen, both a frontal view and corneal section at the time of evisceration or enucleation. The to show the depth of the ulcer. Particular comeal specimen itself was divided in half, one attention was paid to the size, depth, and half for culture. The other half was fixed in for- edges of the ulcer and the greatest diameter malin and sent to the Department of was measured for future comparison. The Pathology, Institute of Ophthalmology, presence and height of a hypopyon were London, for histology. recorded, together with other evidence of All the clinical details, results of progress, anterior chamber reaction or iris and lens and results of Gram stain and culture were involvement. entered into a specially prepared database and analysed at the International Centre for Eye Health in London. CORNEAL SPECIMENS FOR CULTURE AND MICROSCOPY Local anaesthetic without preservative was Results instilled (oxybuprocaine eyedrops) and a A total of 207 consecutive cases presenting sterile Kimura spatula was used to scrape the with suppurative keratitis were studied. base and edges of the ulcer. This material was Records were incomplete or specimens missing inoculated onto, firstly, a 'chocolate' (lysed in eight cases, so that 199 cases have been blood) agar plate and, secondly, a Sabouraud analysed. The demographic and social agar slope. If the ulcer had obvious fungal characteristics of the 199 patients with features when viewed under the slit-lamp, an suppurative keratitis are shown in Table 1. The additional Sabouraud slope in a bijou bottle mean age was 36-3 years, the youngest patient was inoculated to be sent directly to the was 1 year old while the oldest was aged 80 mycologist in London. years. The majority (69-3%) of the patients Further comeal scrapings were then taken for were male. Twenty five different occupations smears on at least two glass microscope slides. were represented, the largest groups were stu- These were labelled and allowed to dry in air. dents/teachers (20s1%) and traders (19 6%). Slides were fixed in 95% methanol for 5 Agricultural workers, an occupational group minutes and then stained in the clinic with usually thought to be at particular risk of sup- routine Gram's method. One slide was exam- purative keratitis, accounted for 16 1 % of the ined under X 10, X40, and finally under x 100 patients. (oil immersion) lenses, to identify bacteria, An eye injury during the previous 3 months hyphae, and other fungal elements. The Gram was reported by 77 (39-2%) of the patients. stain findings were recorded in the patient's The most common causes of eye trauma were study record and in the hospital notes. wood, sticks, and twigs (18 patients), other On the basis of the Gram stain, the organ- vegetation (10 patients), and stones, sand, and isms seen were classified into six categories: dirt (17 patients). No eye injury was reported Gram positive cocci in clumps or clusters by 122 (60-8%) patients. (staphylococci), Gram positive cocci in chains or diplococci (streptococci), Gram positive rods, Gram negative cocci, Gram MICROBIOLOGICAL DIAGNOSIS BASED ON negative rods, and fungal hyphae or yeast CULTURE forms. One or more pathogens were cultured from the 1026 Hagan, Wright, Newman, Dolin, Johnson Table 1 Demographic and social characteristics of 199 Table 3 Combinations ofpathogens cultured patients with suppurative keratitis Pathogen cultured Patients % Br J Ophthalmol: first published as 10.1136/bjo.79.11.1024 on 1 November 1995. Downloaded from http://bjo.bmj.com/ on June 14, 2020 by guest. Protected by copyright. Charactenstics No Gram +ve bacteria only 25 12-6 Age (years) Gram -ve bacteria only 22 11 1 <15 33 16-6 Fungi only 56 28-1 15-29 51 25-6 Gram +ve and -ve bacteria only 3 1-5 30-44 51 25-6 Gram +ve bacteria and fungi only 4 2-0 45+ 64 32-2 Gram -ve bacteria and fungi only 2 1-0 Sex Gram +ve and -ve bacteria and 1 fungus 1 05 Male 138 69-3 Gram +ve and -ve bacteria and 2 fungi 1 0-5 Female 61 307 Nothing cultured 85 42-7 Place of residence Accra region 160 80-4 Total 199 100-0 Volta region 7 3-5 Central region 12 6-0 Eastern region 16 Not known 4 20 bacteria were cultured from 34 patients, Gram Occupation Farming 32 16*1 negative bacteria from 29 patients, while fungi Wood/stoneworker 14 70 were grown from a total of 64 patients. Metal worker 7 35 In one patient four different orgarnsms were Other factory workers 12 Student/teacher 40 260o cultured, including two fungi: Staphylococcus epi- Trader 39 19-6 dermidis, Pseudomonas aeruginosa, Lasiodiplodia Driver 14 Other 20 1700 theobromiae, and Dichotomophthoropsis species. In Retired/unemployed 18 91 another case three organisms were identified, Not known 3 1,5 Vibrio metschnikovii, an a haemolytic strepto- Eye injury within previous 3 months No 122 60-8 coccus, and a filamentous fungus which did not Yes 77 39-2 survive in transit for further identification. Wood, stick, twig 18 No pathogen was cultured for 85 (42-7%) Other vegetable matter 10 Stone, sand, dirt 17 patients, although 52 of these patients had a Other foreign body 32 pathogen identified by microscopic examina- tion of a smear taken from their corneas. corneal smears of 114 patients (57 3%). Overall, no pathogen was found either by Shown in Table 2 is the range of bac-teria and microscopy or by culture for 33 (16-6%) fungi cultured. The most common organisms patients. isolated were Fusarium species, Pseiudomonas To determine why no pathogen was found for aeruginosa, and Staphylococcus epide rmidis. A these 33 patients, possible explanatory factors single pathogen only was cultured fErom 103 were examined. No difference was found patients while two or more different p athogens between the 33 patients and the remaining 166 were cultured from 11 patients. Cattegorising patients for whom a pathogen was found with the cultured pathogens according to their regard to use of eye medicines before assess- Gram staining (Table 3) shows Grana positive ment at the clinic, interval between onset of symptoms and attendance at the clinic, or diam- Table 2 Organisms cultured from corneal scral >ing taken eter of epithelial defect or diameter of infiltrate firom 199 patients with suppurative keratitis in the affected eye. However, the groups did dif- Number ojf ulcers fer significantly in relation to the quality of the Organism culturingjDossnve smear collected from the cornea for mnicrobio- Gram positive bacteria 8 logical diagnosis. When the quality of the smear Streptococcus pneumoniae was categorised arbitrarily as poor, adequate, or Streptococcus sp 3 Enterococcus faecalis 1 good 19 of the 33 (57T6%) patients with no Corynebacterium sp Staphylococcus aureus 4 pathogen found had a poor quality smear com- Staphylococcus epidermidis 14 pared with 55 of the remaining 166 (34 0%) Propionibactenium acnes 1 patients (x2, p=003). Total 3 Gram negative bacteria Moraxella sp 4 Haemophilus influenzae 1 Neisseria gonorrhoeae 2 COMPARISON OF CULTURE BASED DIAGNOSIS Neisseria sp Pseudomonas aeruginosa 16 AND GRAM STAINING BASED DIAGNOSIS Pseudomonas sp 1 2 The microbiological diagnosis based on culture Enterobacter cloacae Vibrio metschnikovii was compared with microscopic examination of Alcaligenes sp 1 smears taken directly from the cornea (Table Fungi Total 29 4). Of the 34 patients for whom Gram positive Fusarium solani 6 bacteria were cultured, 17 were identified Fusarium dimerum Fusarium sp 21 correctly by direct microscopy in the ophthal- Aspergi;usfumigatus 1 mic clinic (sensitivity=50%), while for the Aspergillusflavus 5 remaining 17 cases Gram positive bacteria were Aspergillus terreus Aspergillus sp 3 either not detected or incorrectly identified by Pseudalkscheria boydii Cladosponum sp I 4 microscopy. Of the 29 patients from whom Lasiodiplodia theobromiae 6 Gram negative bacteria were cultured, 13 were Trichosporon capitatum Nigrospora sp correctly identified by microscopy in the oph- Candida parapsilosis thalmic clinic (sensitivity=45%), and of the 64 Curwlanum fallax Acremonium sp patients from whom fungi were cultured, 34 Phoma sp were correctly identified by microscopy (sensi- Dichotomophthoropsis sp 2 tivity=53%). Unidentified fungi Total 65 Microscopy in the ophthalmic clinic often identified bacteria which were not cultured. Causes of suppurative keratitis in Ghana 1027 Table 4 Evaluation of microscopy of Gram stained corneal smears when undertaken by ophthalmic clinic staff and by a medical microbiologist, compared with culture results Br J Ophthalmol: first published as 10.1136/bjo.79.11.1024 on 1 November 1995. Downloaded from http://bjo.bmj.com/ on June 14, 2020 by guest. Protected by copyright. Ophthalmology clinic staff Medical microbiologist Seen by microscopy Sensitivity Specificity Seen by microscopy Sensitivity Specificity Pathogen Culture No (No) (%) (%O) (No) (%) (%) Gram +ve Yes 34 17 50 76 18 53 87 bacteria No 165 40 21 Streptococcus Yes 12 8 (GPDC, GPC, chains) 67 76 9 (GPDC, GPC, chains) 75 88 No 187 45 23 Staphylococcus Yes 18 2 (GPC, chains) 11 85 4 (GPC, chains) 22 93 No 181 27 12 Gram-ve Yes 29 13 45 84 22 76 92 bacteria No 170 28 14 Pseudomonas Yes 17 6 (GNR) 35 85 12 (GNR) 71 92 No 182 28 14 Fungi Yes 64 34 (hyphae) 53 87 51 (hyphae) 80 93 No 135 17 10 GPDC=Gram positive diplococci; GPC=Gram positive cocci; GNR=Gram negative rods. Out of the 165 patients for whom no Gram determined and 116 (58&3%) patients had positive bacteria were cultured, 40 were identi- their treatment changed following the micro- fied as having Gram positive bacteria by direct biological assessment (Table 5). Treatment microscopy at the ophthalmic clinic (speci- was started or changed empirically in 18 ficity=76%). Similarly, of the 170 patients for patients, while in total 122 (61303%) had their whom no Gram negative bacteria were cultured, treatment either determined or altered based 28 were identified by microscopy as having on the results of the microbiological diagnosis. Gram negative bacteria (specificity= 84%), Of these, 87 had their treatment and of the 135 patients for whom fungi were determined/changed solely on the basis of the not cultured, 17 were identified by microscopy direct microscopic examination. The micro- as having fungal hyphae by microscopy scopic diagnosis for these 87 patients was com- (specificity=87%). pared with diagnosis based on culture (Table To further explore the sensitivity and 6). Ten of 11 Gram negative organisms, seven specificity of microscopy based diagnosis by of 11 Gram negative organisms, and 29 of 33 ophthalmic clinic staff, the Gram stained fungal hyphae were correctly diagnosed by smears were sent to Worthing, UK and exam- ophthalmic clinic staff using direct ined by an experienced medical microbiologist. microscopy. The corresponding sensitivity of This was undertaken to differentiate the use- direct microscopy was 91%, 65%, and 88% fulness of microscopy under optimum condi- respectively for Gram positive organisms, tions compared with that at the busy Gram negative organisms, and fungi. ophthalmic clinic in Accra. The sensitivity of microscopy based diagnosis improved when undertaken by the microbiologist (right hand Conclusion side of Table 4). For fungi and Gram negative Before this study started, it had been estimated bacteria, microscopy was able to identify that fungi comprised approximately 10% of correctly 80% and 76% of culture positive cases of suppurative corneal ulcer in Accra. In cases, respectively. The specificity and false fact, they constituted either alone or in combi- positive rate for microscopy based diagnosis nation, over half (56%) of those from whom a also improved when undertaken by a micro- culture result was obtained - as high a propor- biologist. This seemed to reflect the micro- tion as has been recorded anywhere so far. biologist's ability to differentiate between From the published reports, it is apparent that particulate matter and pathogens and between there is a gradual increase in the proportion of Gram positive and Gram negative bacteria, suppurative keratitis due to fungus as one goes and the longer time available for the from higher latitudes in the northen hemi- microbiologist to examine the slide. sphere towards the equator. There is also a general tendency for a greater number of fungal species to be isolated and identified in IMPACT OF MICROBIOLOGICAL DIAGNOSIS ON tropical latitudes, although some published TREATMENT studies are much more comprehensive than Twenty (10.1%) patients had their treatment others. Accra is not only at 5.50 latitude north and hot, but also in general has very Table S Impact of microbiological assessment on treatment humid conditions which may be expected to encourage the growth of filamentous fungi in Reasons for change in treatment the environment. Treatment received Change of treatment Microscopy Culture Microscopy and Other Fusarium was the commonest genus of before assessment after assessment results only results only culture results reasons fungus identified in Ghana. In this respect, No No 13 Ghana resembles the United States rather than Yes 20 14 3 1 2 India, Nepal, or Bangladesh, where Aspergillus Yes No 44 Yes 116 72 17 11 16 has so far been the commonest genus reported. Not known No 2 This is further evidence for geographical varia- Yes 4 1 1 2 0 Total No 59 tion in the distribution of fungi pathogenic for Yes 140 87 21 14 18 the eye, which in turn influences the choice of treatment. It is also interesting that the 1028 Hagan, Wright, Newman, Dolin, J7ohnson Table 6 Evaluation of microscopy, when undertaken by ophthalmic clinic staff, and antifungals.2 The prevalence of previous treat- culture results for 87 patients for whom microscopy findings directly resulted in ment was similar in culture negative and commencement or change of treatment Br J Ophthalmol: first published as 10.1136/bjo.79.11.1024 on 1 November 1995. Downloaded from http://bjo.bmj.com/ on June 14, 2020 by guest. Protected by copyright. culture positive cases in the present study. It is Pathogen Culture No Seen by microscopy (No) Sensttttvty (%) Spectfic4y (%) possible that some of the culture negative cases Gram +ve bacteria Yes 11 10 91 68 could be accounted for by anaerobic organisms No 76 24 or by Acanthamoeba. Appropriate culture Gram -ve bacteria Yes 11 7 65 76 methods have therefore been introduced and a No 76 18 Fungi Yes 33 29 88 72 search is being made for these organisms. No 54 15 The next stages of this programme will be to determine the sensitivity of the fungal isolates to simple antifungal substances which could predominant organism in south Florida has potentially be made available at a reasonable changed over time, from Fusarium solani price in isolated situations in tropical countries; between 1959 and 1977, giving way to to decide the optimum antibacterial and anti- Fusarium oxysporum between 1982 and 1992.13 fungal agents for the organisms isolated; and to From the present report and other published establish trials of the simple agents with reports, it is clear that it is predominantly optimum regimes for prophylaxis after injury filamentous fungi, not yeasts, that cause and of early treatment. infection in the eye in tropical climates. The identification by Yvonne Clayton of fungal isolates is grate- Agricultural occupation was uncommonly fully acknowledged. We thank Alison McCartney for the histo- associated with suppurative infection in logical examinations, and Christine Ntim-Ampousah, Vera Ofri-Darko, Agatha Aboe, Akua Brobbey, Elizabeth Ofori- Ghana, contrary to reports from other regions. Mante, and Francis Codjoe for their assistance with collection This was true also when fungal infections were of clinical data and identification of organisms. This study was supported by grants from the British Council considered in isolation from bacterial infec- for Prevention of Blindness. tions. Of 63 proved fungal cases, 12 (19%) 1 Thomas PA. Mycotic keratitis - an underestimated were students, 12 traders, and only eight mycosis. J Med Vet Mycol 1994; 32: 235-56. (12-7%) were farmers. 2 Liesegang TJ, Forster RK. Spectrum of microbial keratitis results show in South Florida. AmJ Ophthalmol 1980; 90: 38-47. The microbiologist's 3 Williams G, McClellan K, Billson F. Suppurative keratitis microscopy is particularly useful for the identi- in Bangladesh: the value of Gram stain in planning man- fication of fungi and Gram negative bacteria agement. Int Ophthalmol 1991; 15: 131-5. 4 Dunlop AA, Wright ED, Howlader SA, Nazrul I, Husain R, (sensitivity=80% and 76%, specificity=93% McClellan K, et al. Suppurative comeal ulceration in and 92% respectively). The reduced value of Bangladesh. A study of 142 cases examining the micro- biological diagnosis, clinical and epidemiological features the technique when undertaken by staff at the of bacterial and fungal keratitis. Aust NZ J Ophthalmol ophthalmic clinic in Ghana indicates that clinic 1994; 22: 105-10. 5 Thomas PA. Keratomycosis (mycotic keratitis). Bailliire's staff required more thorough training and dinical tropical medicine and communicable diseases. retraining than was thought to be necessary at London: Bailliere, 1989; 4: 269-86. 6 Upadhyay MP, Karmachanya PCD, Koirala S, Tuladhar the start of this study. For microscopy to have NR, Bryan LE, Smolin G, et al. Epidemiologic character- its maximal application, the slide must istics, predisposing factors, and etiologic diagnosis of comeal ulceration in Nepal. Am J Ophthalmol 1991; 111: evidently be read by a person trained and 92-9. 7 Coster DJ, Wilhelmus K, Peacock J, Jones BR. Suppurative experienced in microscopy and the necessary keratitis in London. IVth Congress of the European time must be available in a busy clinic Society of Ophthalmology. Royal Society of Medicine False positive cultures tended to be of International Congress and Symposium Series No 40. London, 1981: 395-8. Gram positive species, such as Staphylococcus 8 Asbell P, Stenson S. Ulcerative keratitis. Survey of 30 years epidermidis, which may be contaminants from laboratory experience. Arch Ophthalmol 1982; 100: 77-80. the normal flora of the tear film and eyelids. 9 Carmichael TR, Wolpert M, Koornhof HJ. Corneal ulcera- What is the reason for no culture being tion at an urban African hospital. BrJ7 Ophthalmol 1985; 69: 920-6. obtained on 85 scrapings, and no pathogens by 10 Ormerod LD. Causations and management of microbial either microscopy or culture in 33? It appears keratitis in subtropical Africa. Ophthalmology 1987; 94: 1662-8. that the material obtained was too small in 11 Gugnani HC, Talwar RS, Njoku-Obi ANU, Kodilinye HC. some of these scrapings, although emphasis Mycotic keratitis in Nigeria: a study of 21 cases. Br J Ophthalmol 1976; 60: 607-13. was continually placed on sufficiently vigorous 12 Williams G, Billson F, Husain R, Howlader SA, Islam N, scraping in the training and review sessions. McClellan K. Microbiological diagnosis of suppurative keratitis in Bangladesh. Br J Ophthalmol 1987:71: Forty five per cent of cases in south Florida 315-21. were culture negative, and this was attributed 13 Rosa RH, Miller D, Alfonso EC. The changing spectrum of fungal keratitis in South Florida. Ophthalmology 1994; to partial previous treatment with antibiotics or 101: 1005-13.