Patients and methods

The medical records of patients presenting with infectious corneal ulcers to the eye casualty department of Queen Alexandra Hospital in Portsmouth between January 1997 and December 2003 were reviewed.

A corneal ulcer was defined as a suppurative corneal infiltrate associated with an overlying epithelial defect. The widest diameter was used to determine ulcer size and grouped accordingly into small (<2 mm), medium (2–4 mm) or large (>4 mm). Central ulcers are those located within the pupilary area. The area between the pupilary margin and the limbus was evenly divided into the concentric areas of paracentral, midperipheral and marginal areas.

Corneal scrapes were routinely collected from all patients with corneal ulceration for the microbiology assessment unless a clearly diagnostic clinical pattern of ulcer was present. The corneal scrapes were inoculated onto blood agar, chocolate agar, fastidious anaerobe agar and brain–heart infusion broth for isolation of bacteria; Sabouraud’s dextrose agar for isolation of fungi; and non-nutrient E. coli-enriched agar for isolation of Acanthamoeba. Selective media, such as Lowenstein–Jensen for Nocardia and Mycobacteria were used if clinically indicated.

A positive isolate was defined as a growth along the line of inoculation on solid media or growth in the liquid media that was confirmed by microscopy. Isolates were subjected to in vitro antimicrobial sensitivity testing against the commonly used antibiotics using the disc diffusion method.

The diagnosis of viral ulcers relied in the majority of cases on the characteristic clinical pattern of viral corneal ulcers. In patients with a suspicious clinical pattern, corneal samples were assessed using immuno-fluorescence.

The follow-up duration was the number of days elapsed between patient’s first and the medically recommended last visit on the completion of ulcer resolution. The best corrected presenting visual acuity, using the Snellen chart, was compared with the best corrected visual acuity on the last visit. Patients with impaired visual acuity due to reasons other than the presenting corneal ulceration and patients who did not complete the follow-up were excluded from the visual acuity assessment. Patients were considered to have poor outcome if they lost one or more of the Snellen lines or if a major complication occurred, including central corneal scarring, irrespective of visual acuity changes.

Chi-square and Fisher’s exact test were used for the analyses of categorical data. Cox regression was used to evaluate factors affecting follow-up duration. The 95% CI limits were provided and p value ⩽0.05 was considered significant. SPSS software (v11.0) for Windows was used for statistical analysis.

Results

Epidemiological characteristics

A total of 1786 patients (3.3%) out of a total 53 341 casualty attendees presented with infectious corneal ulcer between January 1997 and December 2003, with an annual rate of 478/100 000. The average age of patients was 45 years, with a predominance of women (974 (54.5%)). The average number of patients seen per year was 255. A simple linear regression indicated a significant linear upward trend (p = 0.04). There was a significant seasonal variation in the number of patients seen (chi-square = 26.5, p = 0.005) with two seasonal peaks in January and July. The general demographic criteria are shown in table 1.

Predisposing factors

The majority of patients (80.9%) had at least one risk factor. The average age of patients varied according to the associated risk factor. The average ages in CL-wearers, patients with eye trauma and patients with ocular diseases were 35, 48 and 52 years, respectively. CL-wearing was the main risk factor in 554 patients (31%). The majority (353, 63.7%) of CL-wearers were women. Soft CL were used by 534 patients (96.4%), while rigid gas permeable and hard CL were used by 13 (2.3%) and 7 (1.3%) patients, respectively. The wearing modality of 51 soft CL-wearers (9.6%) was not documented. Other risk factors, including ocular diseases (424, 23.7%), ocular trauma (115, 6.4%) and systemic diseases (84, 4.7%), are presented in table 2. Mixed risk factors have been identified in 268 patients (15%); 87 (32.5%) of them were on regular medications, including topical steroids (42, 15.7%), topical antibiotics (33, 12.3%), topical anaesthetics (7, 2.6%) and immunosuppresants (5, 1.9%).

Clinical characteristics

Infectious corneal ulcers involved the left, right and both eyes simultaneously in 919 (51.5%), 820 (45.9%) and 47 (2.6%) patients, respectively. The majority of ulcers were single (1591, 89%) and small (1645, 92.1%). Medium- and large-sized ulcers occurred in 100 (5.6%) and 41 (2.3%) patients, respectively. There was a predominance of mid-peripheral localisations (37.7%), followed by marginal (24.3%), central (23.6%) and paracentral (14.3%) zones. Ulcer location varied significantly between CL-wearers and non-wearers (chi-square = 38.2, p<0.0001). One or more of ocular inflammatory signs and complications occurred in 287 patients (16.1%) including anterior uveitis 139 (48.4%), hypopyon 46 (16%), corneal neovascularisation 36 (12.5%), corneal scar 35 (12.2%), corneal abscess 18 (6.3%), endophthalmitis 9 (3.1%), severe corneal thinning or perforation 8 (2.8%), and sclerokeratitis 5 (1.7%).

Microbiological characteristics

In 532 patients (29.7%) corneal ulcers were diagnosed and treated according to the presenting clinical features, 442 (83.1%) of them were treated with antiviral medications, while the remaining 90 (16.9%) ulcers were treated with topical antibiotics. Corneal scrapes from 1254 patients grew positive cultures in 800 patients (63.8%) as shown in table 3.

Pure bacterial growth was detected in 683 (54.5%) cultures with the predominance of single bacterial species (522, 76.4%). The majority of the isolated bacterial organisms were Gram-positive (696, 71.1%) with a predominance of Staphylococcusepidermidis (310, 31.7%). Gram-negative bacterial isolates accounted for 283 (28.9%) of the total 979 bacterial isolates with predominance of P aeruginosa (117, 12%) as presented in table 4.

Ciprofloxacin was the most frequently used topical medication in 692 (88.9%) of the total 778 patients with bacterial keratitis. Only 2% of all bacterial isolates were resistant to ciprofloxacin and no Gram-negative isolates were resistant to gentamicin.

The profile of Gram-positive organisms showed no difference between CL-wearers and non-wearers (chi-square = 12.6, p = 0.125); however, the distribution of Gram-negative bacteria was significantly dependent on CL-wearing (chi-square = 27.7, p<0.0001).

Visual acuity changes

The presenting visual acuity ranged from 6/4 to hand movement. The majority (790, 44.2%) presented with a visual acuity equal to or better than 6/6. The presenting visual acuity was 6/9, 6/12, 6/18 and 6/24, or worse in 386 (21.6%), 318 (17.8%), 145 (8.1%) and 147 (8.3%) patients, respectively. The presenting visual acuity was significantly dependent on ulcer size (chi-square = 142.6, p<0.001) and ulcer location (chi-square = 324.2, p<0.001). Visual acuity, after ulcer resolution, was improved, worsened or showed no change in 217 (12.1%), 34 (1.9%) and 1310 (73.4%) patients, respectively. Visual acuity change was not assessed in 225 patients (12.5%): 58 visited once, 95 did not complete the follow-up and 72 had impaired vision due to factors other than corneal ulceration.

The average age of the 34 patients who had poor visual outcome was 50 years, with male predominance (18, 52.9%). All of them had at least one predisposing factor and CL-wearing accounted for 41.2%. The majority (16, 47.1%) presented with visual acuity 6/18 or worse. All patients completed the follow-up in average 21.5 days. Twenty patients (58.9%) had large or medium-sized ulcers and in 27 patients (79.4%) ulcers were centrally located. Bacterial growth was detected in 17 patients (50%) with the predominance of P aeruginosa (32.3%). Herpes simplex virus, Acanthamoeba and Candida albicans were isolated from 12 (35.3%), three (8.8%) and two (5.9%) patients, respectively. Poor visual outcome was significantly dependent on ulcer location and size, presenting visual acuity, ocular complications, infection with P aeruginosa or Acanthamoeba (p<0.0001), age (p = 0.0005) and presence of risk factor (p = 0.0249).

Attendance and follow-up

Fifty-eight patients (3.2%) attended once (95% CI 2.5 to 4.2) while the majority of patients (1728, 96.8%) repeatedly sought medical help; 1633 patients (91.4%) completed the follow-up till ulcer resolution (95% CI 90 to 94) and 95 patients (5.3%) did not complete the follow-up (95% CI 4 to 7).

A highly significant relationship was found between follow-up completion and the presence of risk factors, presenting visual acuity, type of infection, ulcer location (p⩽0.0001), ulcer size (p = 0.006), age (p = 0.012) and occurrence of complications (p = 0.056). The follow-up duration ranged from 7 to 42 days with an average of 11.5 days. Cox-regression analysis for factors affecting the follow-up duration in 1633 patients who completed follow-up is presented in table 5.

Old age, being female, presence of more than one predisposing factor, poor presenting visual acuity, large and centrally located ulcers, and having P aeruginosa or Acanthamoeba keratitis were significantly more likely to prolong the duration of follow-up (p<0.001). The lower the value of the hazard ratio, the more likely the patient to be old, female or have a severe form of keratitis.

Discussion

The current study included 1786 patients with infectious corneal ulcers who attended the eye casualty department at Queen Alexandra Hospital, Portsmouth, one of the busiest hospitals in south England. This large population is likely to reflect the exact rate of infectious corneal ulcers.

The annual number of patients with infectious corneal ulcers averaged 255 cases, representing 3.3% of the total ophthalmic attendees. In New Zealand7 infectious keratitis accounted for 2% of total attendees. In the Hong Kong study,4 223 patients were recruited in 17 months and in France6 291 patients were examined in 20 months. The relatively high rate of infectious keratitis reported in our series could be due to the long period of our study. Furthermore, unlike the French study6 that involved only patients with pure bacterial keratitis, our study included all types of infectious ulcers.

The average age of patients was 45 years, exactly the same age as reported in the New Zealand study,7 but was higher than age averages reported in France,6 Taiwan8 and Switzerland5 (39, 40.7 and 44.3 years, respectively). The average age varied according to risk factor type. CL-related keratitis occurred in young patients at an average age of 35 years compared with 48 and 52 years in patients with corneal trauma and ocular disease, respectively. Different studies reported comparable young mean age in CL-related keratitis.6910 The average age also varied significantly according to infection type. In bacterial and viral keratitis, the average age was 47 years compared with 32 years for Acanthamoeba keratitis. This low average age could be due to the tendency of Acanthamoeba to occur in CL-wearers, who tended to be young. Our results showed that 81.8% of Acanthamoeba patients were CL-wearers.

Female preponderance in our study may be because the majority of CL-wearers were females. Women accounted for 63.7% and 50.4% of CL-wearers and non-wearers, respectively. McNally et al10 reported that 70% of lens-related infiltrative events occurred in women.

Our results showed seasonal variation in infectious keratitis in patients among whom CL-wearing was the main risk factor. Despite the fact that there is no clear, obvious or direct correlation has been found between CL-wearing and seasonal variation in the literature, it has been suggested that lens wear has an at least additive effect that increases the possibility of infectious keratitis.11

At least one predisposing factor was detected in 80.9% of patients. This is lower than the results of other keratitis studies: 88%,7 90.6%6 and 97%12. Lens wear, associated ocular disease and corneal trauma were the main risk factors that matched the findings of other studies.567121314

Thirty-one per cent of our patients wore CL. However, CL-wearing may account for up to 56% of patients9 compared with only 0.6% in places where CL-wearing was unpopular.15 A declining trend in the rate of lens-related keratitis since the late 1990s has been noticed, following the introduction of fluoroquinolones.16 The relatively low rate of lens-related keratitis in our study could be because only patients with true ulceration were included and not cases with stromal infiltrations covered with intact epithelium, which are frequently associated with CL-wearing and resolve upon discontinuation of CL-wearing. The inclusion of frank ulcers avoided giving false overestimation of the actual figures of lens-related keratitis.6 Soft lenses were used by 96.3% of patients compared with reported values of 92.5%,17 89.4%6 and 53%.9 Hard and rigid gas-permeable lenses were used by 3.6% of patients compared with reported values of 10.6%6 and 28.4%.18

It is agreed among different authors that ocular surface diseases predispose to infectious keratitis.46 History of ocular diseases accounted for 23.7% of the total population; this was comparable to a previous report of 21%.6

Gram-positive organisms accounted for 71.1% of the proved bacterial growth. The predominance of Gram-positive bacteria in our study is comparable with other results.67111920 Gram-negative bacteria accounted for 28.9% of the total positive isolates, higher than the 17% reported in France6 and 18.2% in Japan,20 but less than the 40% reported in the USA.12P aeruginosa was the main Gram-negative bacterium isolated from 12% of all isolates. This predominance of P aeruginosa was comparable to the values of 14%12 and 21.1%21 reported in other keratitis studies.

Our results showed that Gram-positive bacteria were the main bacteria isolated from 61.9% of CL-wearers. The predominance of Gram-positive bacteria accounted for 67%6 and 82.9%18 of CL-related keratitis. However, these results are at variance with the results of other investigators49 who showed predominance of Gram-negative bacteria in CL-wearers that accounted for up to 70% of patients.17

In our study S. epidermidis was the main bacteria isolated from patients with infectious keratitis. S. epidermidis is the most frequently isolated bacterium from ocular flora.22 The predominance of coagulase-negative staphylococci isolated from keratitis patients has been reported.5613 In addition, S. epidermidis strains from endophthalmitis, corneal ulcers and conjunctivitis represented 34.1%, 33.9% and 46% of the reported oxacilin resistance, respectively.23 Genetic differences between pathogenic strains of S. epidermidis isolated from keratitis and endophthalmitis and those constituting the normal commensal conjunctival flora have been reported,24 with the suggestion that commensal strains can evolve into pathogenic ones.25

Anterior uveitis was a common feature in 48.4% of patients and 16% of them had hypopyon either at presentation or during the course of the infection. Wong et al7 reported anterior uveitis in 59% of cases and 20% exhibited a hypopyon at presentation.

Acanthamoeba keratitis was diagnosed in 1.6% of lens-related infectious keratitis. The spectrum of lens-related Acanthamoeba keratitis ranged from 0.49%21 to 4.7%.26Acanthamoeba and P aeruginosa infections were associated with severe keratitis, particularly in CL-wearers where it eventually led to poor visual outcome. In our series 1.9% of patients had poor visual outcome compared with 5%6 and 5.8%5 reported in other studies.

As expected, centrally located and large corneal ulcers were associated with a high rate of poor presenting visual acuity and poor vision outcome; these findings matched those in other studies.713 Large and central ulcers are commonly complicated with endophthalmitis and needed surgical treatment with penetrating keratoplasty.13

The mean time of follow-up period, which reflects the duration of corneal ulcer resolution, was 11.5 days. Suchecki et al27 reported a mean time of 1.74 weeks (12.2 days) needed to complete resolution for all corneal infiltrates. Two weeks14 and 14.7 days7 were reported as the average time needed for complete healing of corneal ulcers. Unsurprisingly, the follow-up time of patients was significantly prolonged in elderly patients and in patients with multiple risk factors, compromised corneas, poor presenting visual acuity, resistant infections, and large and centrally located ulcers.7