British Journal of Ophthalmology, 1986, 70, 199-201

Management of bacterial corneal ulcersR MASKE,' J C HILL,' AND S P OLIVER2

From the Departments of'Ophthalmology and 2Medical Microbiology, Groote Schuur Hospital and Universityof Cape Town, Republic ofSouth Africa

SUMMARY A prospective microbiological study of 48 patients with corneal ulcers due to bacterialinfection was performed. Positive cultures of corneal ulcer samples were obtained in 60% of allpatients; about half of these patients had received antimicrobial treatment prior to sampling. Arelatively high incidence of Staphylococcus epidermidis was isolated from ulcer patients (27%)compared with normal controls (10%). Gram stains of ulcer samples were positive for organisms inonly 27% of all patients and were not considered useful in determining initial therapy in this series.We concluded that treatment should be started with a broad combination of antibiotics whileawaiting the culture results.

Despite the easy access of the cornea to therapeuticagents the treatment of bacterial corneal ulcersremains a problem. There are basically two ap-proaches to the management. The specific therapyadvocated by Jones' is most widely accepted and isbased on the examination of corneal scrapings withculture identification of the offending organism andtreatment of the ulcer according to its antibioticsensitivity. Alternatively the broad therapy deve-loped by Baum2 is based on the prevalence oforganisms known to produce disease in a particulargeographic location. A broader treatment regimenaccording to the known sensitivities of the most likelyorganisms is given.Both approaches require a knowledge of the

causative organisms and their sensitivities. This paperpresents the findings of a microbiological study ofpatients with bacterial corneal ulcers referred toGroote Schuur Hospital for treatment. Peripheralstaphylococcal ulcers have been specifically excludedfrom the study.

Patients and methods

During the period January 1983 to March 1984, 48patients were seen at Groote Schuur Hospital withclinically diagnosed bacterial corneal ulcers. Allcases of uninfected erosions, viral keratitis, fungalkeratitis, and peripheral staphyloccal infection havebeen excluded from this study. Patients' ages rangedCorrespondence to Dr R Maske, Department of Ophthalmology,University of Cape Town, Medical School, Observatory 7925, CapeTown, Republic of South Africa.

between 8 and 80 years with an average of 50 years;69% were males and 31% females.On admission of the patients to hospital corneal

samples were taken for Gram stain and culture. Thecornea was anaesthetised with 0-4% oxybuprocainesolution without preservative (Minims Benoxinate).In each case samples were taken from the base andleading edge of the ulcer by means of sterile calciumalginate swabs,3 care being taken to avoid touchingthe conjunctiva and lids. The material was spreadover a localised area on precleaned glass slides forimmediate Gram stain and microscopic examination.Further samples were taken for direct inoculationinto cooked meat medium, boiled blood agar, andSabouraud's agar for culture identification of organ-isms. Corneal and conjunctival swabs were taken forsimilar Gram stain and culture from 39 controlpatients with clinically uninfected eyes admitted forroutine surgery such as lens extraction and squintcorrection.Our treatment regimen followed that advocated by

other authors.'3 The initial treatment was based onthe results of the corneal smears. All patients re-ceived 1% atropine drops twice daily. Gram-positivecocci were treated with subconjunctival cephaman-dole and hourly topical 10% cephamandole drops,and Gram-negative bacilli were treated with subcon-junctival gentamicin and hourly topical 3% gen-tamicin drops. If no organism was identified bothcephamandole and gentamicin were used. The sub-sequent management of all ulcers was determined bythe culture and sensitivity results as well as theclinical response to treatment.

199

R Maske, J C Hill, and S P Oliver

Results

GRAM STAIN RESULTSPus cells were present on the smears taken from 38patients (79%). However, in all the remainingpatients who showed no pus cells on the smear therewas clinical evidence of infective keratitis, with afrank hypopyon in all but one of these cases.Organisms were identified on the smears taken from13 patients (27%), seven patients showing Gram-positive cocci and six Gram-negative bacilli.

CULTURE RESULTSPositive corneal cultures were obtained from 29patients (60%). Gram-positive cocci were most com-monly isolated, but Gram-negative bacilli and mixedbacterial ulcers were also identified (Table 1). Fifteenof these patients had received antibiotic therapy(chloramphenicol 14; neosporin 1) for a variableperiod before being referred to us and, therefore,before their corneal smears were taken for culture.Of the 19 patients with negative corneal cultures

eight had received antimicrobial therapy beforebeing referred to us and culturing (chloramphenicol

Table 1 Organisms culturedfrom corneal ulcers

Number of PriorOrganism isolates therapy*

Gram-positive cocci: 20 12Staph. aureus 4 2Staph. epidermidis 9 6Str. pneumoniae 6 3Str. viridans 1 1

Gram-negative bacilli: 5 1Pseudomonas aeruginosa 3 1Moraxella species 1 -

Haemophilus influenza 1 -

Mixed infections: 4 2Staph. epidermidis/diphtheroids 4 2

*See text for details.

Table 2 Gram stain and culture correlation

Gram stain Culture No. patients

Gram and culture correspond (8 patients)Gram-positive cocci: Str. pneumoniae 2

Staph. epidermidis 2Gram-negative bacilli: Pseudomonas 2

aeruginosaMoraxella 1speciesHaemophilus 1influenza

Gram and culture different (5 patients)Gram-positive cocci: No growth 3Gram-negative bacilli: No growth 1

Str. viridans 1

Table 3 Organisms isolatedfrom 39 controlpatients

Number of isolates

Organism Conjunctiva (%) Cornea (%)

No organisms 33 87Staph. epidermidis 49 10Staph. aureus 13 0Bacillus species 5 3

5; framycetin 1; tobramycin 1; sulphacetamide 1).Of the 13 patients with positive Gram stains

organisms were cultured from nine; eight of thesewere appropriate to the Gram stain and the other onewas different from what the Gram stain showed(Table 2). From the remaining four patients noorganisms were cultured.Although we cultured organisms from 60% of

patients, not all positive cultures were necessarilysignificant. There was a relatively high incidence ofStaphylococcus epidermidis (27% of all patients),which is a normal resident of conjunctiva and lids.4To clarify the role of Staph. epidermidis we tookcorneal and conjunctival swabs from 39 controlpatients with clinically uninfected eyes admitted forroutine surgery.

NORMAL CONTROLSOrganisms were isolated from 67% of conjunctivalswabs but only 13% of corneal swabs taken from thesame eye of 39 control patients (Table 3). There was ahigh incidence of Staph. epidermidis cultured fromthe conjunctival swabs, thus confirming its residentialstatus in the conjunctiva. Most of the corneal organ-isms cultured were also Staph. epidermidis; in allcases the organisms cultured from the corneal swabswere also found in the conjunctival cultures, indicat-ing that these organisms were present in the pre-corneal tear film.

Discussion

The successful management of bacterial cornealulcers is based on prompt identification of the causa-tive organism and effective treatment with an appro-priate antibiotic.'35 Unfortunately the Gram stainand culture results are not always positive. Wecultured organisms from 60% of patients with cor-neal ulcers, and this compares favourably with theresults of other authors, who have reported inci-dences of positive cultures ranging between 53% and73%*.A number of factors may influence the culture

results. Usually only a small amount of material isavailable, giving a light growth of organisms. Topicalanaesthetic agents are known to have antimicrobial

200

Management ofbacterial corneal ulcers

activity and their use may lead to false negativeculture results.8We instilled preservative-free MinimsBenoxinate immediately before taking samples forculture, as it does not interfere significantly with theisolation of organisms from corneas.' This madethe sampling procedure more comfortable for ourpatients, enabling us to obtain adequate sampleswithout touching the lids or conjunctiva.Treatment of ulcers with antibiotics prior to cor-

neal culturing may also reduce the number of positiveresults. Almost half (23 patients) of all our cornealulcer patients received topical antibiotics beforesmears were taken for culture. However, it is signifi-cant that about half of the patients with positivecultures had been treated with antibiotics prior toculturing. Thus previous treatment does not necess-arily preclude positive culture results and should notbe a reason for avoiding sampling.

In this study we have followed the type of regimenadvocated by Jones' in which the initial therapy wasbased on the results of the Gram stain and treatmentwas subsequently adjusted according to the cultureand sensitivity results. Our Gram stain results weredisappointing because organisms were identified inonly 27% of all patients, and subsequent cultureconfirmed the identity of these organisms in only57% of these patients. The Gram stain was thereforenot very useful in determining the appropriate initialtherapy in our series. In one patient a differentorganism was cultured. On the Jones regimen thispatient received inappropriate medication until thetime that the culture report was available. Since itoften takes 48 hours before the culture results areknown, we now prefer to ignore the Gram stain andstart treatment with a broader combination regimenwhile awaiting the culture results.5 Treatment maysubsequently be modified according to the sensitivityof the organism as well as the patient's response totreatment.A relatively high incidence of Staph. epidermidis

was isolated from our corneal ulcer patients (27%).This organism is a normal resident of conjunctiva andlids,4 and its role in ocular disease has been thesubject of much controversy recently.79 10 It is knownto be pathogenic in some immunologically com-promised patients."'3 We found that in all normalcontrol cases with positive corneal cultures the sameorganism was isolated from the conjunctival culturesfrom the same eye. These organisms were thereforepresent in the precorneal tear film and could eithercolonise or contaminate an area of corneal ulceration

sufficiently to be detected on the corneal culture.'Unfortunately this does not answer the question ofthe pathogenicity of Staph. epidermidis. However, itmay be pertinent that a higher incidence of Staph.epidermidis was cultured from the patients withcorneal ulcers (27%) than from the normal controls(10%). This suggests that some of these ulcers weredue to exogenous Staph. epidermidis or alternativelythat patients having this organism resident are moreprone to infection. Mahajan et al.7 12 have shown thatapproximately one-third of isolates of Staph. epider-midis taken from various types of ocular infectionsshowed dermonecrotoxic activity and were thereforeregarded as pathogenic. Although Staph. epidermidismay not be of importance in all ulcers, it cannot bedisregarded, and it would seem prudent to treat suchan infection actively.

We thank Professor A A Forder for encouragement and guidanceand Miss B Robinson for microbiological assitance.

References

1 Jones DB. Initial therapy of suspected microbial corneal ulcers.11. Specific antibiotic therapy based on corneal smears. SurvOphthalmol 1979; 24: 97,105-16.

2 Baum JL. Initial therapy of suspected microbial corneal ulcers. 1.Broad antibiotic therapy based on prevalence of organisms. SurvOphthalmol 1979; 24: 97-105.

3 Pettit TH. Management of bacterial corneal ulcers. In: LeopoldIH, Burns RP, eds. Symposium on ocular therapy. New York:Wiley, 1976; 8: 57-65.

4 Seal DV, Barrett SP, McGill JI. Aetiology and treatment ofacute bacterial infection of the external eye. Br J Ophthalmol1982; 66: 357-60.

5 Chaudhuri PR, Godfrey B. Treatment of bacterial corneal ulcerswith concentrated antibiotic eye drops. Trans Ophthalmol SocUK 1982; 102: 11-4.

6 Asbell P. Stenson S. Ulcerative keratitis. Survey of 30 yearslaboratory experience. Arch Ophthalmol 1982; 100: 77-80.

7 Mahajan VM. Acute bacterial infections of the eye: theiraetiology and treatment. Br J Ophthalmol 1983; 67: 191-4.

8 Badenoch PR, Coster DJ. Antimicrobial activity of topicalanaesthetic preparations. Br J Ophthalmol 1982; 66: 364-7.

9 Valenton MJ, Okumoto M. Toxin producing strains ofStaphylococcus epidermidis. Arch Ophthalmol 1973; 89: 186-9.

10 Smolin G, Okumoto M. Staphylococcal blepharitis. Arch Oph-thalmol 1977; 95: 812-6.

11 Forse RA, Dixon C, Bernard K, Martinez I, McLean APH,Meakins JL. Staphylococcus epidermidis-an important path-ogen. Surgery 1979; 86: 507-14.

12 Mahajan VM, Reddy TN, Agarwal LP. Toxigenic strains ofStaphylococcus epidermidis and their experimental cornealpathogenicity in rabbits. Int Ophthalmol 1982; 5: 155-61.

13 Valenton MJ, Brubaker RF, Allen HF. Staphylococcus epider-midis (albus) endophthalmitis: report of two cases after cataractextraction. Arch Ophthalmol 1973; 89: 94-6.

Acceptedfor-publication 8July 1985.

201