Review of epidemiological features, microbiological diagnosis and treatment outcome of microbial keratitis: Experience of over a decade

Purpose: To review the epidemiological characteristics, microbiological profile, and treatment outcome of patients with suspected microbial keratitis. Materials and Methods: Retrospective analysis of a non-comparative series from the database was done. All the patients presenting with corneal stromal infiltrate underwent standard microbiologic evaluation of their corneal scrapings, and smear and culture-guided antimicrobial therapy. Results: Out of 5897 suspected cases of microbial keratitis 3563 (60.4%) were culture-proven (bacterial – 1849, 51.9%; fungal – 1360, 38.2%; Acanthamoeba – 86, 2.4%; mixed – 268, 7.5%). Patients with agriculture-based activities were at 1.33 times (CI 1.16–1.51) greater risk of developing microbial keratitis and patients with ocular trauma were 5.33 times (CI 6.41–6.44) more likely to develop microbial keratitis. Potassium hydroxide with calcofluor white was most sensitive for detecting fungi (90.6%) and Acanthamoeba (84.0%) in corneal scrapings, however, Gram stain had a low sensitivity of 56.6% in detection of bacteria. Majority of the bacterial infections were caused by Staphylococcus epidermidis (42.3%) and Fusarium species (36.6%) was the leading cause of fungal infections. A significantly larger number of patients (691/1360, 50.8%) with fungal keratitis required surgical intervention compared to bacterial (799/1849, 43.2%) and Acanthamoeba (15/86, 17.4%) keratitis. Corneal healed scar was achieved in 75.5%, 64.8%, and 90.0% of patients with bacterial, fungal, and Acanthamoeba keratitis respectively. Conclusions: While diagnostic and treatment modalities are well in place the final outcome is suboptimal in fungal keratitis. With more effective treatment available for bacterial and Acanthamoeba keratitis, the treatment of fungal keratitis is truly a challenge.

At presentation to the cornea services of this institute, information pertaining to demographic features, duration of symptoms, risk factors and occupational status was documented for every suspected case of infectious keratitis according to a detailed protocol. Cornea evaluation was carried out by a cornea specialist using a slit-lamp biomicroscope and Þ ndings were recorded in a predesigned format. Detailed diagrammatic documentation of the ulcer was done and recorded on a daily basis. Treatment regimen, response to treatment and Þ nal outcome were recorded in all cases.
Following clinical examination, patients were subjected to microbiological investigations as per the institutional protocol described earlier. [9,10] The bacterial and fungal isolates were identiÞ ed up to the species level using standard microbiological procedures. 11 The smear and culture results were recorded in the predesigned format along with clinical details and captured in the corneal ulcer database which is maintained systematically. All analysis projected in this study is derived from this database.
The standard protocol used for the treatment of our patients is described in detail in an earlier publication which reported microbial keratitis in an elderly population seen at this institute from February 1991 until June 1995. [10] The treatment protocol has remained unchanged since then for bacterial and fungal keratitis, however, we have adopted combined therapy with polyhexamethylene biguanide (0.02%) and chlorhexidine (0.02%) for Acanthamoeba keratitis since August 1996. [12] Surgical mode of treatment included tissue adhesive application with bandage contact lens, penetrating keratoplasty, evisceration, whenever applicable. Treatment outcome at the end of three months or at the completion of treatment (whichever was earlier) was considered for analysis.

Statistical analysis
Student's t test was applied to compare the mean values of demographic factors such as age. The chi square test was used for comparison of proportions. The odds ratio (OR) with 95% conÞ dence interval (CI) was employed to assess the relative risk of patients with trauma and agriculture-related occupation developing microbial keratitis.

Results
Of the 5897 clinically suspected cases of infectious keratitis, 4087 (69.3%) were males and 1810 (30.7%) were females, the overall male to female ratio of patients being 2.25:1. Laboratory evidence of microbial infection was established in 3563 (60.4%) of 5897 cases whose corneal scrapings were subjected for smears and culture. The mean (± standard deviation) age was 41.20 (± 20.36) years in patients with bacterial keratitis (1849, 51.9%), 30.90 (± 15.28) years in patients with fungal keratitis (1360, 38.2%), and 34.45 (± 12.54) years in patients with Acanthamoeba keratitis (86, 2.4%), indicating a relatively increased occurrence of corneal infections (irrespective of the etiological agent) in the middle age group. The seasonal variation in the occurrence of all (including mixed) bacterial, fungal and Acanthamoeba keratitis is as depicted in Fig. 1.
Unilateral ulcer cases included 1789 right eyes and 1737 left eyes. Thirty-seven patients had bilateral infection accounting for 3600 aff ected eyes. Since both eyes of patients with bilateral infection revealed identical organisms, the occupational status, possible risk factors, duration of symptoms, prior medication, and laboratory parameters were analyzed taking into account 3563 patients and not eyes.
The occupations of patients [ Table 1] were classiÞ ed as outdoor (agriculture and manual labor), and indoor (desk job and household). More number of patients with fungal, Acanthamoeba (pure cultures) and polymicrobial keratitis (bacteria and fungus; bacteria and parasite) were found to be involved in agriculture-related activities (P < 0.001) as compared to other occupations; this feature was not evident in patients with pure bacterial keratitis and in cases where fungus and Acanthamoeba coexisted. Odds ratio (OR) revealed that patients involved in agriculture-based activities were 1.33 times (CI 1.16-1.51) at a greater risk of developing microbial keratitis.
The potential predisposing ocular factors identified in patients are shown in Table 2. Between the three etiological groups (pure cultures), the association of trauma was more pronounced for fungal and parasitic keratitis as compared to bacterial (P < 0.001). Overall, patients with ocular trauma were 5.33 times (CI 6.41-6.44) at a greater risk of developing microbial keratitis.
Patients with outdoor occupation had higher prevalence of keratitis due to trauma as compared to the patients engaged indoors. This observation was signiÞ cant for bacterial (P < 0.001), fungal (P < 0.001) and Acanthamoeba (P = 0.02) keratitis when all culture-positive trauma and non-trauma cases were considered. In keratitis of pure or polymicrobial origin, physical agents were the most frequent sources of corneal injury than the other two (P < 0.001) as depicted in Fig. 2. Among the systemic factors documented in 296 patients, diabetes mellitus was more frequently noted in keratitis of both pure and polymicrobial etiology, accounting for 69.2% (205/296) cases.
Among the 3563 patients, 1945 (54.6%) were treated with antimicrobial agents and corticosteroids topically elsewhere,    prior to their presentation to our cornea services [ Table 3]. When retrospectively analyzed it was observed that in 945 (48.6%) patients the antimicrobial agents received were partly or completely in agreement with the type of the microbial agent (bacterial or fungal) causing the infection as proven by culture. Most patients, however, had received the medications in less than optimum dosage.
Overall, greater number of patients had sought medical help at our institute with duration of symptoms less than one month (2977) than those with symptoms longer than one month (405) as shown in Fig. 3 Table 4. On analysis of matching smear and culture results, Gram stain was accurate in only 45.7% of the corneal scrapings from 2115 patients with bacterial keratitis (pure and mixed). Among the 2334 culture-negative cases, smears were positive for microorganisms in 739 (31.7%) cases revealing bacteria in 417 (17.9%), fungus in 298 (12.8%), Acanthamoeba in 19 (0.8%) and both bacteria and fungus in Þ ve (0.2%) eyes. These cases being culture-negative were not analyzed in this study.
Of 3563 cases of microbial keratitis, 3295 (92.5%) revealed pure growth of either bacteria (1849, 51.9%), fungi (1360, 38.2%) or Acanthamoeba (86, 2.4%). Polymicrobial infection was seen in 268 (7.5%) cases. Of the 37 cases that presented with bilateral infection, 34 cases demonstrated pure bacterial growth, two had pure fungal growth, and one had mixed infection of both bacteria and fungus, both eyes of each of these patients revealing similar organisms. Vol. 57 No. 4 More than one bacterium (two or more) was isolated from 350 cases resulting in 2511 bacterial isolates. Among the bacterial isolates, 2062 (82.1%) were gram-positive and 449 (17.9%) were gram-negative. The diff erent bacterial and fungal species isolated are listed in Table 5 and 6 respectively. Propionebacterium (19, 0.8%) and Peptostreptococcus (seven, 0.3%) species were the only anaerobes recovered in this series. The antibiotic susceptibility data of the bacterial isolates is beyond the scope of this study and is published elsewhere. [13,14] Overall, 1648 fungal isolates were recovered from culture of corneal scrapings (50 patients had more than one isolate). Of these, 1635 (99.2%) were molds and 13 (0.8%) were yeasts.
All patients were started on medical therapy initially, however, 46.6% of the patients required surgical intervention as shown in Table 7. Overall treatment outcome in bacterial, fungal, and Acanthamoeba keratitis patients is shown in Table 8. Significantly more number of patients required surgical treatment in fungal keratitis compared to bacterial and Acanthamoeba keratitis.

Discussion
A variety of factors determine clinical outcome in microbial keratitis and the epidemiological patterns vary from one country to the other and in diff erent geographical areas in the same country. A comprehensive data is important to develop appropriate diagnostic and therapeutic strategies. This study reports the experience with 3563 culture-positive non-viral microbial keratitis patients based in southern India. The data reported here is expected to be useful in all areas of the world where fungal keratitis is relatively more prevalent and is commonly considered in the diff erential diagnosis of microbial keratitis.
The male preponderance in this series was observed not only in the overall clinically suspected cases of microbial keratitis but also in culture-proven cases of microbial keratitis (male:female::2.25:1, 2.24:1 respectively). Though both sexes develop corneal ulcers more commonly in the middle decades of life, a signiÞ cant male preponderance has been reported by most previous studies [5,15] including those in children [2,16] and elderly patients. [2,11] Considering the predominant predisposing factor of trauma in all types of microbial keratitis (bacterial -46.6%, fungal -81.9%, Acanthamoeba -95.5%) the probable reason for male preponderance is obvious. Ocular trauma was signiÞ cantly more associated with outdoor occupation in this series.
More than half of the patients with culture-proven microbial keratitis (54.6%) had visited a physician prior to presentation at this institute and nearly half (48.6%, Table 3) of them had received antimicrobial agents that were appropriate, albeit on lower dosage, for the microbial agent involved. Therefore, we believe that despite the patient being on prior antimicrobial therapy, microbiological investigation may succeed in establishing etiological diagnosis in at least 50% of the patients. Traditional medicine or home remedy was used by only 0.4% of our patients compared to 37.3% of the patients in the study from Madurai. [5] The urban location of our institute in contrast to the semi-urban location of the institute at Madurai may account for this diff erence. While use of plant extracts has been reported from rural Malawi, Africa by Courtright et al. [17] , it is fortunately not common in areas undergoing urbanization.
It is interesting to note that a majority of our patients presented within one month of onset of symptoms, 42% of whom came within one week. This indicates easy availability of transport to patients and is in contrast to the situation in other developing countries such as Nepal [2] where 19.3% of the patients took longer than one month to reach the hospital for treatment. Transport facilities and access to healthcare systems are important issues in the developing countries and our analysis points at optimum availability in the area catered by this institute.
Direct microscopic examination of corneal scrapings provides rapid diagnosis and forms the basis for instituting initial antimicrobial therapy which may be modiÞ ed later according to culture reports. [18] An accurate smear diagnosis therefore becomes important in achieving optimum treatment outcome. The detection of fungi and Acanthamoeba was much higher in the smears than it was for bacteria [ Table 4]. The detection rate for bacteria (Gram stain) was reduced by 10.9% when a correlation of the presence of similar bacteria in smears and cultures was made. We recently analyzed the utility of  Gram stain in the diagnosis of early and advanced bacterial keratitis wherein the sensitivity was found to be 36.0% and 40.0% respectively. [19] The low sensitivity was att ributed by us to the use of antibiotics prior to presentation at this institute by nearly 50% of the patients. The sensitivity of Gram stain in the diagnosis of bacterial keratitis, as reported by other authors (Asbell et al. [20] -67%, Dunlop et al. [21] -62%), is close to the overall sensitivity noted in this study (56.6%) which dropped on correlation of presence of similar bacteria in smears and cultures (45.7%).
Microorganisms were isolated in 60.4% of the 5897 cases of presumed microbial keratitis. This Þ gure is close to many other reports [4,5] but is lower than the reports from Nepal [2] (80%) and from Bangladesh (81.7%). [21] The protocol of culture techniques followed in this study and the procedure of sample inoculation directly in the clinic leaves virtually no scope for role of laboratory-related reasons for low yield in culture. Patient-related causes such as prior antimicrobial therapy probably have a signiÞ cant role to play, as has been suggested by Srinivasan et al. [5] A majority of our patients (3295/3563, 92.5%) had corneal infection by a single agent, the most common being bacterial (1849/3563, 51.9%). Bacterial keratitides were predominantly caused by gram-positive bacteria. However, unlike other studies from Asia [2,5] and Africa [22] where infections by Streptococcus pneumoniae were most common; in our study, Staphylococcus epidermidis-related bacterial keratitis predominated. A review of literature showed that most of the studies from developed countries such as the USA [1,15,20] (except southern USA) and Australia [23] listed S. epidermidis or coagulase-negative staphylococci as the leading cause of bacterial keratitis. It is not clear whether the tendency to consider S. epidermidis or coagulase-negative staphylococci as a normal commensal of the conjunctiva may have led to underreporting in some of the    studies. Nevertheless, the criteria to determine the signiÞ cance of a positive culture from corneal scrapings appeared similar across most of these studies. Considering the fact that S. epidermidis forms the commonest commensal of the extraocular surfaces, it is highly probable that these organisms invade corneal tissues when compromised by antimicrobial and/ or corticosteroid therapy or trauma. The higher incidence of S. pneumoniae keratitis in Madurai compared to this series remains inexplicable since both these studies are from southern India. The strong association of chronic dacryocystitis with S. penumoniea-related microbial keratitis is well known [24] but the database in this study was not adequate to determine the frequency of concomitant sac pathology in our patients. It is possible that a larger number of patients with dacryocystitis were present in studies with predominant S. pneumoniae infection.
A high prevalence of fungal keratitis caused by Þ lamentous fungi in warmer climates has been widely reported. [1,4,25] All cases (pure and polymicrobial) were considered together in this series; fungi were isolated in 1598 (44.8%) patients, a frequency similar to that reported from Madurai. [5] Some of the fungal isolates could not be deÞ nitely identiÞ ed due to lack of characteristic spores [ Table 6] in the medium used at our center for culturing fungus (Sabouraud dextrose agar, potato dextrose agar). Diffi culty in speciation of fungi owing to lack of sporulation has been faced by other investigators as well. [1] Att empts were not made in this study to use spore-enhancing media for fungal isolates on a routine basis, which probably would have helped in speciation of some of the unidentiÞ ed isolates.
The overall incidence of Acanthamoeba keratitis (3.3%) was low in this study although the number of affected patients was large (118). Only one patient had worn contact lenses (0.8%). In contrast to the literature from developed countries, where contact lens wear emerges as a great risk factor for developing infectious keratitis, [26,27] it accounted for only 42 out of 3563 (1.2%) cases in this series of which the majority (36/42, 85.7%) were bacterial. No patient among a series of 33 cases of Acanthamoeba keratitis, recently reported from south India, had worn contact lenses. [28] Concomitant infection with bacteria (0.8%) and fungi (0.1%) was quite rare in patients with Acanthamoeba keratitis. Diagnosis based on initial smear examination of corneal scraping was most rewarding in calcoß uor white stained smears by ß uorescence microscopy.
A signiÞ cantly larger number of patients (691/1360, 50.8%) with fungal keratitis required surgical intervention compared to bacterial (799/1849, 43.2%) and Acanthamoeba (15/86, 17.4%) keratitis thus indicating a poor response to treatment in fungal keratitis compared to bacterial and Acanthamoeba keratitis (P < 0.05). This study shows that although bacterial and Acanthamoeba keratitis can be treated eff ectively, the treatment of fungal keratitis remains a challenge.