Corneal foreign bodies: are antiseptics and antibiotics equally effective?

To compare the effect of antiseptics and antibiotics on the occurrence of Infectious Keratitis (IK) secondary to Corneal Foreign Body (CFB) removal. Multicenter retrospective study conducted between June 2020 and June 2022 in patients referred for CFBs and treated with Picloxydine (Group 1) or Tobramycin (Group 2) for 7 days. A follow-up visit was scheduled on Day 3 (D3) and a phone call on D30. The primary outcome measure was the occurrence of IK. 307 patients (300 men) with a mean age of 42.8 (14.8) years were included. The mean (SD) time to consultation was 43.1 (45.6) hours. Picloxydine and Tobramycin were given to 155 and 152 patients. Half of patients (n = 154, 50.2%) were building workers and 209 (68.1%) did not wear eye protections. CFBs were mainly metallic (n = 292, 95.1%). Upon referral, rust was found in 220 patients (72.1%). A burr was used in 119 (38.9%) patients. IK occurred in 15 (4.9%) patients, 8 (5.3%) in Group 1 and 7 (4.5%) in Group 2 (p = 0.797). IK was successfully treated in all cases. Persistent rust was found in 113 patients (36.9%) on D3 without difference between burr or needle use (p = 0.278). On D3, corneal healing was delayed in 154 patients (47.2%), mainly in burr-treated patients (p = 0.003). The mean (SD) work stoppage duration was 0.32 (0.98) days. IK rate was 4.9%. The efficacy of antibiotics and antiseptics was similar on CFB removal. Using a burr was associated with a longer healing time. CFBs had a limited social impact.


INTRODUCTION
Corneal Foreign Bodies (CFBs) are a common ophthalmic disorder worldwide. The main symptoms include red eye, eye pain or discomfort, photophobia, tearing, blepharospasm and blurred vision. Eye protection to prevent CFBs is a long-standing concern [1]. However, despite the development of appropriate protection, CFBs remain common in workers [2]. Although being one of the most common ophthalmic emergencies, many questions on CFBs are still unanswered [3]. First, the social impact of CFBs due to occupational or home exposure in terms of work stoppage has not been fully investigated [4]. Second, there is still no consensus on the best method for CFB removal (use of a burr versus a needle) [5]. Finally, CFBs are associated with an increased risk of external contaminants and subsequent infectious keratitis (IK) that may be associated with an irreversible visual impairment [6,7]. IK has been shown to occur in up to 3% of CFB cases [8,9]. Therefore, most ophthalmologists prescribe topical antibiotics following CFB removal. However, no study supports the use of topical antibiotics in this indication and a body of evidence shows that acquired bacterial resistance is promoted by inadequate antibiotic prescription [10]. To our knowledge, no study has compared the effect of antiseptics and antibiotics on the management of CFBs.
The aim of this study was to compare the effect of antiseptics and antibiotics on the occurrence of IK secondary to CFB removal.

MATERIAL & METHODS
No approval from an ethics committee was needed because it was a retrospective study conducted in accordance with the tenets of the Declaration of Helsinki.
This was a multicenter retrospective study, and patients were included in two separate centers (Nice University Hospital and Cannes Hospital, France) between June 2020 and June 2022.

Patients
The records of all patients referred to the ophthalmology emergency department for superficial CFB over the study period were reviewed. Patients were divided into 2 groups: those treated with Picloxydine eye drops (group 1) and those treated with Tobramycin eye drops (group 2). Both treatments were given four times daily for 7 days. The choice to administer the antibiotic or the antiseptic was at the ophthalmologist's discretion. All patients had a follow-up visit scheduled on Day 3 and received a phone call about 1 month later. In case of IK, a corneal sample was taken for analysis and hourly topical bi-antibiotic therapy (aminoglycoside + quinolone or vancomycin + ceftazidime in more severe cases) was prescribed for 48 h followed by antibiotic decrease (treatment duration of 10-21 days depending on IK severity).
Patients treated with systemic antibiotics during the 30-day follow-up, treated with topical antibiotics or steroid eye drops before referral, with IK upon referral, with multiple CFBs and with an incomplete follow-up (missing data on Day 3 and/or 1 month after CFB occurrence) were excluded.

Data recorded
The following data were recorded: • Patient characteristics: age, gender, occupation, time between CFB occurrence and consultation, previous history of CFB over the last 6 months, type of accident (i.e., domestic or at work), wearing of an eye protection at the time of the accident, chronic systemic or ophthalmological disorders, chronic ophthalmologic treatment, treatment initiated before the consultation (saline solution, artificial tears, vitamin A ointment).
• At the time of the first consultation: CFB type and size, affected corneal area and corneal quadrant (Fig. 1a), use of analgesics, uncorrected visual acuity (UCVA), pain score on a 0-10 numerical scale, anterior chamber response, method used for CFB removal (30-gauge needle or burr), size of the corneal ulcer and rust persistence after CFB removal.
• On Day 3: occurrence of IK, UCVA, eye pain based on the same numerical scale, use of analgesics, rust persistence, anterior chamber response, size of the remaining corneal ulcer, superficial punctate keratitis, corneal scar, total work stoppage duration, use of steroids. A corneal scar was defined as a whitish corneal opacity located within the stroma, in the absence of eye pain, eye redness and epithelial ulceration. Topical steroids were prescribed in case of central or paracentral corneal scar in the absence of a corneal ulcer and in combination with topical antiseptics or antibiotics. Topical steroid drops were initiated at a dosage of 3 drops per day for a total duration that remained at the ophthalmologist's discretion.
• On Day 30, patients received a phone call and were questioned about the occurrence of any complication (red eye, eye pain, blurred vision) and the total work stoppage duration. In case of red eye, eye pain, or persistent blurred vision, a new ophthalmological consultation was scheduled.

Measurement of clinical outcomes
The primary endpoint was the incidence of IK in the tobramycin and picloxydine groups on Day 30. IK was diagnosed based on the ophthalmological examination and consisted in red eye, eye pain, associated with a whitish corneal infiltrate with an overlying corneal ulcer.
Secondary endpoints were the social impact of CFBs in terms of work stoppage, the factors associated with a persistent corneal ulcer on Day 3, and the factors associated with rust persistence on Day 3.

Statistics
Quantitative variables are expressed as a mean (SD) and discrete outcomes as absolute and relative (%) frequencies. Differences between groups were assessed by comparing baseline demographics and the follow-up duration. Normality and heteroskedasticity of continuous data were assessed using a Shapiro-Wilk test and a Levene's test, respectively. Continuous outcomes were compared using ANOVA, Welch ANOVA, or Kruskal-Wallis tests depending on data distribution. Discrete outcomes were compared using a chi-squared or Fisher's exact test. The alpha risk was set at 5% and two-tailed tests were used.

Patient's baseline characteristics
The records of 580 patients with CFBs were reviewed, and 307 patients met the inclusion criteria. Two hundred and fifty-eight patients were excluded due to an incomplete follow-up as shown in Fig. 1b. Patients' characteristics are shown in Table 1. Most patients were men (n = 300, 97.7%) and the mean (SD) age was 42.8 (14.8) years. One hundred and fifty-four (50.2%) patients were building workers and 41 (13.4%) were metal workers. Two hundred and nine (68.1%) patients experienced CFB while they were not wearing any eye protection and CFBs occurred at work in 175 (57.0%) patients. A history of CFB during the last 6 months was found in 32 (10.4%) patients. Upon presentation, the mean (SD) pain score was 5.1 (2.5). The mean UCVA was 0.09 (0.20) logMAR. Most CFBs were metallic (n = 292, 95.1%). The mean (SD) time to consultation was 43.1 (45.6) hours. At baseline, rust was found in 220 (72.1%) patients. CFBs in the presence or in the absence of rust were removed with a burr in 119 (38.9%) patients whereas a 30-G needle was used in 187 (61.1%) patients. As shown in Table 1, both groups were similar at baseline, except for a central corneal involvement and a past history of CFB that were more common in Group 2 (respectively 30.0% versus 20.6%, p = 0.029 and 14.6% versus 6.5%, p = 0.034).

Outcome measures
Of the 307 patients, 155 received Picloxydine (Group 1) and 152 received Tobramycin (Group 2) after CFB removal. The results of the primary and secondary outcomes are shown in Table 2. IK occurred in 15 (4.9%) patients: 8 (5.3%) in Group 1 and 7 (4.5%) in Group 2. No statistical difference was found between both groups (p = 0.797). The mean time to IK occurrence after CFB and after the first consultation was 5.3 days (range: 3-8) and 3.8 days (range: 2-7), respectively. An illustrative case of IK secondary to CFB is shown in Fig. 1c. The IK cases are briefly presented in Table 3. Although all the corneal samples were negative, a bacterial cause of all IK cases was suspected and they were treated accordingly. All patients were successfully treated with a combination of two antibiotics. At the end of the follow-up, only one patient underwent phototherapeutic keratectomy. No corneal graft was required. The results at the end of the follow-up are shown in Fig. 1b.
Of the 307 patients, 113 (36.9%) had residual rust on Day 3. The use of a burr (p = 0.278), the mean time to consultation (p = 0.302) and CFB size (p = 0.241) were not significantly associated with the presence of residual rust on Day 3. In the multivariate analysis (Supplementary Table 1), only the presence of rust upon referral (p < 0.001) and the persistence of rust after CFB removal (p < 0.001) were significantly associated with the persistence of rust on Day 3.
Also, 175 (57.4%) patients experienced a corneal scar on Day 3. The CFB size, the use of a burr and the presence of rust after CFB removal were not associated with corneal scarring. In the multivariate analysis, only non-iron CFBs were significantly associated with the presence of a corneal scar on Day 3.
A corneal ulcer persisted in 154 (47.2%) patients on Day 3. The CFB size and the size of the ulcer after CFB removal were not associated with a persistent corneal ulcer. In the multivariate analysis, the mean time to consultation (p = 0.017), the presence of rust on Day 0 (p = 0.008) and the use of a burr for CFB removal (p = 0.003) were significantly associated with the persistence of a corneal ulcer on Day 3. No patient required the use of an amniotic membrane graft during the follow-up.
The mean work stoppage duration was 0.32 (0.98) days. Ninety-nine (32.4%) patients required the use of analgesics on Day 3. The mean pain score decreased from 5.12 upon referral to 1.11 on Day 3 (p < 0.001).
At one month, 26 (8.8%) patients experienced persistent eye pain and 13 (4.4%) complained of blurred vision. Table 4, this multicenter retrospective study conducted in patients with CFBs was the largest and first one to compare the use of antiseptics and antibiotics. Of interest, we found that IK occurred in 4.9% of cases, with no statistical difference between the antiseptic and antibiotic groups. Despite common belief, the work stoppage duration was short (mainly on the day of consultation), thus limiting the social impact of CFBs. Using a burr was not associated with better rust removal and seemed to increase the risk of persistent corneal ulcer on Day 3.

As shown in
We compared our results with previous CFB studies (Table 4) and we found the highest rate of IK secondary to CFB removal (4.9%). In their study, Smadar et al. and Brissette et al. found an IK rate of 3.2% and 1.6%, respectively [8,9]. This difference could be explained by the small number of patients included in their studies (62 and 60, respectively), i.e., 5 times less patients than in our study. In addition, our study was the first to follow the patients for one month. Brissette et al. have treated their patients with fourth-generation fluoroquinolone that has a different antibacterial spectrum compared to the aminoglycosides we used [8,11]. In the study by Smadar et al., treatment was given at the ophthalmologist's discretion, making a direct comparison impossible [9].
In this study, we used topical aminoglycosides because they have a broad antibacterial spectrum and are active against most IK-related pathogens. Previous studies have found that staphylococci and Pseudomonas aeruginosa were the most prevalent microorganisms found in post-traumatic IK [12]. As shown in Fig. 1d, the anti-bacterial spectrum of aminoglycosides and quinolones is almost similar, except that aminoglycosides are often effective against methicillin-resistant staphylococci while quinolones are not [13]. Conversely, quinolones are effective against chlamydia while aminoglycosides are not.
According to the European recommendations, fourthgeneration fluoroquinolones should be avoided as first-line therapy due to an increased rate of acquired resistances worldwide [14]. There are many complex mechanisms of resistances, and some are still poorly understood [15]. This increased resistance to quinolones also impacts the ocular surface flora [16]. Therefore, fluoroquinolones should only be prescribed in dedicated clinical situations [17]. In France, topical thirdgeneration fluoroquinolones are only used to treat IK and should not be prescribed as an antibiotic prophylaxis.
Aminoglycosides are also associated with acquired resistance. In their study, Budzinskaya et al. have given a prophylactic topical aminoglycoside antibiotic (tobramycin) in 20 patients for 8 days following intravitreal injections [10]. Tobramycin administration was associated with a 4.5-fold increase in conjunctival bacterial resistance (from 20% to 90%) among conjunctival staphylococci.
Topical azithromycin is a newly available broad-spectrum antibiotic in France, belonging to the macrolide family, that acts against Gram-positive and Gram-negative bacteria, including intracellular bacteria [18]. However, azithromycin is not active against Pseudomonas aeruginosa due to intrinsic resistances (Fig. 1d) which explains why this antibiotic is not prescribed for the management of CFBs [19,20]. Topical azithromycin is given twice daily for 3 days due to its long half-life [21]. This reduced instillation may increase patient compliance, although pain at instillation is a common side effect [22].
Picloxydine acts by creating an electrostatic bond between positively charged antiseptic molecules and negatively charged cell walls. Picloxydine has a destructive effect on the bacterial wall and plasma membrane, leading to cellular integrity loss and release of intracellular components [23]. Picloxydine appears interesting for the management of CFBs for 3 reasons: (i) Picloxydine has a broad antibacterial spectrum (Table 4) and also has an activity against Streptococcus unlike fluoroquinolones and aminoglycosides; [23,24] (ii) an acquired resistance to picloxydine seems to be rare, Pyrogov et al have found a resistance rate to  [25,26] (iii) picloxydine has been shown to have an antiparasitic activity against amoebas that could be responsible for co-infection in IK secondary to CFB [27]. Taken together, these data suggest that picloxydine could be an alternative to topical antibiotics in CFBs since it has an adequate antibacterial spectrum and is associated with lower rates of acquired resistances. Further studies are needed to confirm this assumption.
Contrary to previous studies that have not identified any difference between the use of a burr and a needle, we found an increased risk of persistent corneal ulcer on Day 3 when a burr was used [3]. This could be explained by a potentially deeper stroma removal when a burr was used compared to a needle [28]. As shown in a previous study, the rust ring induces a local corneal toxicity [29]. Thus unsurprisingly, a longer time to consultation and the presence of rust on Day 0 were associated with longer corneal healing. Despite common belief and in accordance with Alexander et al., the social impact of CFBs appears low with only 0.32 days of work stoppage in developed countries [30].

Study limitations
Due to its retrospective design, the rate of patients lost to followup was high (44.5%) in this initial cohort (Fig. 1b). This is not surprising in a study investigating CFBs, because most patients are young and active men (Table 4) who do not want to lose working time for a follow-up consultation. As shown in Table 3, the mean pain score was reduced to 1.1 on Day 3 and pain is usually the first symptom leading to consultation. The broad objective of this study was to compare topical antiseptics with topical antibiotics. However, the use, recommendations and availability of topical antibiotics and antiseptics differ greatly from a country to another. This variability is also highlighted by the method of extraction of the foreign bodies. A burr is not always available, although we have shown that it may delay the corneal healing. These variations from a country to another lead any study comparison challenging. Table 4 summarizes the current studies on corneal foreign bodies and highlights the local habits.
In our study, the choice to prescribe antiseptic or antibiotic eye drops was left at the ophthalmologist's discretion. This could have introduced an unconscious bias, with more severe cases treated with antibiotics. However, patients' characteristics were identical in both groups ( Table 1). The social impact of CFBs has not been fully investigated. Although the work stoppage duration was reduced, the visual impact was not evaluated. In this retrospective real-life study, we did not record the best-corrected visual acuity (BCVA) but only the UCVA. Indeed, measuring the BCVA in an emergency context, with a painful and watery eye is challenging and time consuming. In addition, CFBs may significantly alter the ocular surface by inducing irregular astigmatism. At 1 month, 4.4% of patients complained of blurred vision.

CONCLUSION
In conclusion, we found an occurrence rate of IK secondary to CFB removal of 4.9%. The efficacy of Picloxydine and Tobramycin appeared similar. Due to increasing acquired bacterial resistances to antibiotics, more studies investigating the use of topical antiseptics are needed. Using a burr was associated with an increased risk of persistent corneal ulceration. Finally, the social impact in terms of work stoppage appeared anecdotal.

SUMMARY
What was known before Corneal foreign bodies are among the most frequent ophthalmological emergencies. Infectious keratitis is the most feared complication of superficial corneal foreign body. Antibiotics are widely prescribed following corneal foreign body removal and no study compared topical antibiotics versus antiseptics.

What this study adds
There is no significant difference in the rate of infectious keratitis between the antiseptic group and the antibiotic group. The social impact is limited with a majority of patients working the day after the corneal foreign body removal. Compared to traditional needles, the use of a burr does not remove better the rust and is associated with a corneal healing delay at Day 3.

DATA AVAILABILITY
The data that support the findings of this study are not openly available due to reasons of sensitivity human data and are available from the corresponding author upon reasonable request.