Discussion

Corneal opacity is the fifth leading cause of blindness globally, with an estimated 1.5–2.0 million cases of unilateral blindness annually.8 Ocular trauma and corneal ulceration are significant causes of corneal blindness that are often under-reported, which accounts for up to 5% of all blindness.9 10 Capturing epidemiological data for MK, of which CLAK is a subgroup, is difficult because data are reported under the term “corneal blindness” which comprises a range of traumatic, infectious, inflammatory and inherited conditions.9 This leads to difficulty measuring the true burden of MK, which is likely to be underestimated because the current WHO guidelines define ‘blindness’ as a VA of <3/60 corrected in the better eye and MK often causes disability that is significant but fall short of this standard,9 hence highlighting an ongoing unchecked burden on human health.8 11

The pattern of MK aetiologies varies by region, with bacterial predominance in developed countries and fungal predominance in developing countries.12–15 The highest proportions of bacterial aetiology (over 85%) have been reported in Europe, North America, Australia and Oceania, where the incidence of CL wear are also the highest.16–21 MK costs the US healthcare system an estimated 175 million dollars in direct health expenditures and approximately 70 million dollars in Medicare and Medicaid-related costs.9 However, no such data exist for the developing countries but it is reasonable to conclude that these costs are magnified in the poorest populations due to poor access to healthcare, delayed presentation, under dosage of empirical antibiotics, inappropriate use of steroids and impracticalities of obtaining cultures, and they often occur in the patients’ most productive years.22 23

In the UK, many ophthalmology units share a common system of managing CLAK regardless of severity or the size of epithelial defect, whereby patients are started on empirical treatment with topical antibiotics and a follow-up is arranged in 48 hours to review for improvement or deterioration. Additionally, corneal scrapes are performed for culture and sensitivities for any epithelial defects 1 mm or more in size. With a continuously growing population in London where Moorfields Eye Hospital services are based, there is a sharp rise in demand for eye casualty services where a majority of these patients first present to a medical professional for treatment.24 Previous studies show that corneal presentations account for over 18% of eye casualty presentations, and up to 50% of MK are secondary to CL wear.2 25 This further increases the burden of disease on a resource-limited National Health Service and our study provides practitioners with evidence on safe approaches to managing CLAK or MK to optimise clinical efficiency and channel resources to patients who require them most.

First, our analysis on the risk factors of CLAK are corroborated with previous studies showing similar trends with extended-wear CL, poor hygiene and prolonged duration of wear among the most significant risk factors for developing MK.26–28

It is generally understood that the pathogenesis of CLAK is complex and multifactorial, and a plausible aetiology is related to a reduction of tear exchange during blinking in combination with tear stagnation under the CL, resulting in adherence and accumulation of microbes on the cornea, and reduced corneal epithelial cell desquamation, and alteration of tear fluid biochemistry.29 30

Second, our subgroup analysis of the outcomes in individual CLAK grades show that grades 1 and 2 CLAKs can be classified as low risk as they are predicted to have excellent outcomes with 96% of the grade 1s being discharged at their first appointment or within 48 hours, and 94.1% of the grade 2s were discharged in 2 weeks. Most importantly, the patients discharged in both groups have had a similar average duration of treatment with topical moxifloxacin monotherapy, emphasising that the prognosis of these patients is similar. All the patients that remained under review had presented with atypical features. We describe a typical CLAK as a corneal lesion with classical features of a focal epithelial defect with an underlying subepithelial or stromal infiltrate and an atypical CLAK where additional features of multiple infiltrates, perineual infiltrates, dendritiform lesions, satellite/feathery infiltrate with associated pigment, or endothelial plaques are found that may be suggestive of fungal or acanthamoeba infections. Due to the low number of grade 3 patients, it is difficult to derive useful statistical analysis but the treatment regime in this group of patients is estimated to be higher than average. Therefore, there is a need to review current management protocols in managing CLAKs throughout the UK as low-risk patients (typical grade 1 and 2 CLAK) comprise the majority of patients, and this may also apply globally in other developed countries.

Our results support the safe use of fluoroquinolone monotherapy as first-line treatment which is proven by recent UK studies showing an increasing trend of resistance against penicillin in Gram-positive and Gram-negative bacteria but a maintained moderate and high susceptibility to fluoroquinolone, respectively.17 31 32 Moorfields Eye Hospital’s local recommended empirical regime for CLAKs have been modified to hourly preservative-free moxifloxacin 0.5% for 48 hours, then hourly during the day for 3 days, and four times daily for 7 days. In particular, our findings suggest that patients who present with typical grades 1 and 2 CLAK do not require review in 48 hours and can be discharged on the recommended empirical regime, and patients with atypical features will require follow-up with further diagnostics, additional treatment and monitoring. Furthermore, we need to relook at our threshold for performing corneal scrapes as given our evidence of favourable outcomes of low-risk typical grade 1 and 2 CLAKs (up to 2.0 mm) on fluoroquinolone monotherapy, we should consider performing corneal scrapes for typical lesions with EDs larger than 2.0 mm and have a lower threshold to scrape in the presence of atypical features with the awareness of generally low yields on culture positivity which can vary by region (UK 33.0%–54.0%, Taiwan 49.3%, India 69% and Portuguese 38.4%).17 A recent study from Liverpool comparing the use of corneal impression membrane (CIM) and corneal scrapes found significantly higher bacterial isolate rates from CIM (65.2% vs 19.4%) with significant increases in the isolations of a number of species, including S. aureus (2.4%–11.3%), Serratia (0.5%–1.7%), Streptococcal species (0.7%–6.9%), S. epidermidis (2.1%–26.2%), S. capitis (0.4%–2.6%) and S. warneri (0.3%–1.6%), demonstrating its superiority as an alternative user-dependent method of investigation over corneal scrapes.33 Although CIM had not been implemented at our centre at the time of the study, it has now been introduced at Moorfields which is a simpler and faster procedure to perform compared with corneal scrapes and thus allows allied healthcare staff such as nurses to acquire samples.

Our group recognises several limitations with our study methods and the first of which is with regards to patient population. Moorfields St George’s covers an area of South London and hence our results and local recommendations have to be interpreted cautiously in other geographies and climates where the incidence of atypical pathogens such as fungal and acanthamoeba, and polymicrobial infections differ.34 In view of our findings, a local audit of MK in a similar manner to this study should guide local policy. A retrospective study on CLAK in Southern India found that bacterial infections were the predominant cause at 89.2% and a half of which were positive for pseudomonas on cultures.35 All but one culture of S epidermis was sensitive to ciprofloxacin, demonstrating that similar trends can, however, be found in developing countries although more studies are required to guide antimicrobial management in these regions. The documentation of sensitivities on our electronic patient record system (OpenEyes) was low at 4.35% (1 of 23) and hence we were unable to perform an analysis on the sensitivities of our positive scrapes. Also, the small sample size of grade 3 patients precludes any meaningful statistics and further studies are required to better characterise this group of patients by evaluating the causative microorganisms, sensitivities, treatment duration and prognosis. In addition, the nature of our study design introduces a degree of variability in VA assessment as most patients were not assessed with a pinhole if their initial uncorrected or corrected VA is LogMAR 0.0 or better. Further attempts to collect data for patients who were lost to follow-up from other hospitals can also be made to improve the accuracy of our findings but we recognise the challenge in doing so and our results reflect realistic high rates of DNA in the CLAK patient group.

In conclusion, we present a risk-adjusted management protocol for MK by assessing the outcomes of different gradings of CLAK. Our findings reflect results from a real-world setting with a relatively high number of patients lost to follow-up (25.5%, n=41) and we purport that a majority of patients improved with empirical therapy and voluntarily declined subsequent review with a small number of patients seeking further management in other centres. We recommend that typical grade 1 and 2 CLAK can be discharged with empirical treatment of topical fluoroquinolone with easy access to review at a tertiary centre in the presence of worsening signs or symptoms, and the patients planned for follow-up may be reviewed in 4–5 days depending on access to specialty clinics. In our patient cohort, this would reduce the number of unnecessary reviews in 68.9% (n=111) of all CLAKs and in a healthcare system where demand is steadily on the rise, such measures will go a long way towards optimising the use of resources on patients who need to be seen.