Bringing Eyelid Surgery in Line with International Guidelines Regarding Peri-Operative Antibiotic Agents: A Pilot Study

Patients and Methods

A retrospective audit was undertaken using electronic medical records to establish baseline SSI rates in patients undergoing elective eyelid surgery in a single unit between January 1, 2018 and March 31, 2020. Inclusion criteria were primary closure of wounds and prescription of at least one week of post-operative chloramphenicol ointment. Exclusion criteria included peri-operative use of other topical or systemic antibiotic, evidence of malignancy in any biopsy sample, use of full thickness skin graft or implant, concurrent non-primarily closed wound, e.g., gray line split in anterior lamellar repositioning, dacryocystorhinostomy, and operations in which there is no traditional wound, e.g., isolated use of inverting or everting sutures.

Standard practice at our local institution was to not prescribe pre-operative antibiotic agents in clean, non-contaminated elective periocular surgery. Our practice comprised intra-operative application of chloramphenicol ointment to the closed surgical incision at the end of the procedure prior to the application of an appropriate dressing, followed by a post-operative course of chloramphenicol ointment to the wound for at least one week after the procedure. Local practice was then changed, withholding chloramphenicol ointment both intra-operatively at the end of the procedure and post-operatively for patients undergoing the same caliber of eyelid surgery, and a prospective audit of infection rates undertaken to compare with baseline. Of note, patients for whom the potential sequelae of an infection were deemed to be unacceptably high (for example, patients with corneal grafts, or only one eye with useful vision) were excluded and continued to receive prophylactic chloramphenicol.

Patients were specifically advised to not apply any other ointments to the incision(s), such as paraffin-based lubricants, to avoid the confounding factors associated with ointment application. These could include incision disturbance from handling, introduction of bacteria by touching the incision, and creation of a moist environment on the incision surface, which might potentially encourage bacterial growth or conversely act as a physical barrier to entry of bacteria. Avoidance of other ointments also removed potential confounding effects of any pro- or anti-inflammatory effects of ointment on a wound. Some previous studies have compared infection rates in those receiving topical antibiotic ointment to those receiving topical ointments containing petrolatum or paraffin.

Data were gathered on age of patient at time of surgery, operated eye, grade of operating surgeon, ocular comorbidities, and evidence of infection in their electronic patient records within the three months after their operation. Where electronic records were unclear, written records were examined. Where an infection occurred, specific note was made of suture type, whether sutures were removed, either routinely or to manage a complication, and treatment given. Local ethical approval was granted for this study. χ ² tests were calculated using an online calculator; Shapiro-Wilks, Mann-Whitney U tests, and Kruskal-Wallis independent tests were calculated using SPSS Statistics, Version 25 (IBM Corp, Armonk, NY). Statistical significance was defined as p < 0.05.

Results

A total of 3,136 Medisoft procedures were retrospectively reviewed from the time period of routine chloramphenicol useage, of which 872 eyes of 717 patients met the inclusion criteria of eyelid surgery, primarily closed incisions, at least one week of post-operative chloramphenicol prescribed, and no involvement of malignancy or other exclusion criteria. Fourteen eyes of 13 patients had documented evidence of a post-operative infection (1.6%). Prospectively, 133 eyes of 107 patients were recorded and met the inclusion criteria. Three eyes of three patients had documented evidence of a post-operative infection (2.25%). Table 1 outlines the demographics of both groups.

The only statistically significant differences found were in the level of operating surgeon, with a higher proportion of ophthalmic specialist trainee (OST) operations, and fewer fellow, staff grade and associate specialist doctor (SAS) operations in the prospective, no-antibiotic arm. This was most likely the result of composition of the oculoplastics team during the two time periods and tendency for simple surgeries with primary closure to be more likely to be allocated to trainees. A summary of the characteristics of those patients with evidence of post-operative wound infection are outlined in Table 2.

Median age of patients with infection was 78 years; in those without infection it was 72. There was no statistically significant difference in distribution of ages between those who developed an SSI and those who did not (p = 0.469).

Discussion

Surgical site infections generate a substantial burden for healthcare providers and patients alike, resulting in additional pain, increased number of appointments, possibility for poor incision healing, compromised surgical result and re-operation, and morbidity and mortality. Preventing such infections is thus a matter of great interest, and national and international guidelines from NICE ⁴ and the CDC ⁵ outline measures that should be implemented to minimize the risks of SSI. These are based on SSI classifications that help to risk stratify the likelihood of SSI and the associated risk-benefit balance of interventions such as the use of local or systemic antibiotic prophylaxis. Clean wounds are defined by NICE as “an incision in which no inflammation is encountered in a surgical procedure, without a break in sterile technique, and during which the respiratory, alimentary or genitourinary tracts are not entered,” and they advise against routine antibiotic prophylaxis for these wounds where they have been closed primarily, except where a prosthesis is being implanted. ⁴ Although inter-rater variability has been reported as suboptimal for evaluating surgical site classification, ⁶ the majority of elective oculoplastic operations can be considered clean wounds, and a substantial portion of these will involve primary closure, meaning that antibiotic prophylaxis would not be recommended.

There are certain aspects of peri-ocular surgery that create unusual challenges compared with standard surgical site management. NICE guidelines recommend the application of a “suitable, interactive dressing” for 48 hours post-operatively ⁴ ; in peri-ocular surgery, this may have a substantial functional impact on patients, especially those having bilateral surgery, who have poor vision in the unoperated eye, or have balance or mobility issues whereby loss of binocular single vision may increase the risk of falls. Furthermore, many common oculoplastic procedures such as lateral tarsal strips routinely breach the conjunctiva. Conjunctival colonization is common, with a 2012 study by Halachmi-Eyal et al. ⁷ obtaining bacteria-positive forniceal swab cultures in 40.5% of patients who presented for elective intra-ocular surgery. They found a seasonal variation, with increased rates in spring/summer, although this was an Israeli study, and such variations may not be applicable to the UK population.

NICE guidelines recommend patients shower on the day of surgery, but interestingly, Halachmi-Eyal et al. ⁷ found higher rates of positive cultures in patients who had showered. Conjunctiva could be considered similarly to the mucosa-lined respiratory, alimentary and genitourinary tracts that when breached in otherwise sterile conditions are classified as clean-contaminated wounds; these carry a higher risk of SSI (5%–10%), and antibiotic prophylaxis is recommended for immunocompromised patients. ⁸ In our cohort, 12 of 17 patients with SSI had conjunctival-breaching surgery (ratio of 2.4:1 conjunctival breach: no conjunctival breach) that might support a higher risk of infections in these circumstances, but numbers are too small to draw statistical conclusions. None of these patients were immunocompromised, and thus would not have met NICE criteria for antibiotic prophylaxis, even if we considered conjunctival-breaching surgery as clean-contaminated wounds instead.

It is well-established that prophylactic antibiotics administered pre- or intra-operatively are more effective than those administered post-operatively. ⁹ Patients considered to be at high risk of SSI, or for whom the risks of SSI are deemed unacceptably high, for example patients with poor vision in the non-operated eye, or a corneal graft in the operated one, may receive intravenous peri-operative antibiotics, usually alongside post-operative chloramphenicol ointment. Patients receiving systemic antibiotics were excluded from this study as they might introduce bias to the results due to differing rates of infection, which may have been higher due to the pre-operative patient risk factors, or lower due to the use of intravenous antibiotics. The application of chloramphenicol ointment to the wounds is considered by some clinicians to have additional benefits beyond the antibiotic status, discussed in more detail below, and may be contributing to the ingrained nature of this practice.

Another peri-ocular issue is that of make-up contamination. In the authors' experience, patients frequently report for surgery wearing make-up. Although there is no literature exploring this aspect specifically, we recommend patients be advised not to wear make-up on the day of surgery, and care should be taken to thoroughly remove it prior to surgery to prevent wound contamination when it has been worn.

Despite the level of interest in this subject, to our knowledge there have been no published studies specifically looking at use of prophylactic antibiotic agents on SSI rates for peri-ocular surgery. In 2009, Heal et al. ¹⁰ undertook a double-blinded, randomized, placebo-controlled trial to evaluate the impact of a single dose of topical chloramphenicol to high-risk primarily closed wounds in minor dermatologic surgery. Their baseline infection rates of 8.6% were acknowledged to be higher than published literature, and they speculated the local hot, humid climate may have contributed to this, and the surgeries examined included a high proportion for non-melanoma skin cancer and solar keratosis. It is also noteworthy that the potential pro- or anti-inflammatory effect of ointments are unknown, and this study did not have an arm in which no ointment was applied. They found a statistically significant lower rate of SSIs in those patients who received chloramphenicol at the end of the procedure compared with those who received placebo (6.6% chloramphenicol, 11% control) but this did not reach their pre-defined criteria of a 5% absolute reduction as clinically significant.

A 2015 meta-analysis by Saco et al. ⁸ explored the evidence for antibiotic prophylaxis for clean wounds in dermatologic outpatient procedures. After exclusions, 15 studies were potentially eligible, but only four studies were suitable for meta-analysis. They did not find any statistically significant difference in SSI rates in patients receiving chloramphenicol ointment prophylaxis versus petrolatum/paraffin and recommended cessation of antibiotic prophylaxis for these patients. However, they highlight increased infection risk in patients with diabetes mellitus, skin grafts, flaps on the nose or ear, wedge excisions of lips or ears, basal cell carcinomas and squamous cell carcinomas, and wounds in the groin or below the knee. This meta-analysis formed the basis for our exclusion of patients having surgery involving a malignancy or skin graft. However, we did review our infection rates for such patients during the period of our retrospective (chloramphenicol-treated) arm. Among these patients four of 330 eyes (1.2%) that underwent surgery involving a malignant lesion and two of 36 eyes (5.5%) that received a full thickness skin graft had evidence of SSI. Given the low rates of infection in those undergoing surgery for malignancy, it may be possible to safely withdraw chloramphenicol prophylaxis in this cohort too. Although they are small numbers, the higher rates of infection in patients requiring a skin graft suggest a higher level of caution may be appropriate in these circumstances.

A 2016 Cochrane Review ¹¹ incorporated 14 randomized controlled trials comparing topical antibiotic prophylaxis against no treatment, antiseptic or other topical antibiotics, eight involving general surgical wounds, six in dermatologic. Many were small, low-quality studies at risk of bias. They concluded that use of topical antibiotic prophylaxis probably reduced the absolute risk of SSI by 2%. However, as discussed above, peri-ocular surgery has unique characteristics, which mean that extrapolating such data might not be appropriate.

Influencing factors on surgeons' use of antibiotic prophylaxis despite guidelines to the contrary include lack of awareness of guidelines, adopting habits of institutions or colleagues, fear of litigation, and concerns about the effect on reputation or private practice if a patient has an SSI when antibiotic agents were withheld. Furthermore, some surgeons believe that keeping a surgical scar moist with ointment or encouraging early massage of wounds can help dissolve sutures and improve cosmetic outcome, although this encourages manipulation of the wound by patients and could represent a confounding factor when evaluating infection rates. Some surgeons also routinely prescribe ointment post-operatively for ocular surface lubrication, for example after ptosis surgery, where the risk of worsened dry eye and lagophthalmos are recognized post-operative issues. In our unit, the cost of one tube of chloramphenicol ointment is half that of the least expensive available non-antibiotic lubricating eye ointment. Therefore, for surgeons who would still wish to prescribe an ointment, chloramphenicol would represent the least expensive option. This presents an ethical dilemma when prescribing in a resource-limited healthcare system in an era of major concern regarding antimicrobial resistance. ³

There are several limitations to this study. This work presents the findings from a pilot study undertaken to bring the unit's policies in line with national guidelines. Low pre-existing rates of infection mean that large numbers would be required to identify a statistically significant difference in infection rates. Using a p value <0.05 for significance, assuming a 2% baseline infection rate, to identify a 2% difference would require 1,000 patients per arm. Our baseline infection rate was only 1.6% and identifying smaller differences between the two groups would require prohibitively large numbers for one center. Given the low rates of infection currently, it is unlikely that funding could be obtained to undertake a large multicenter trial to recruit adequate patients and answer this question definitively.

This study relies on electronic medical records; current systems mean that patients may have been treated for SSIs elsewhere, including other hospitals or general practices, and this information may not be known to us or reflected in their electronic medical record. Although this may artificially lower the perceived infection rate, one would expect that such issues would occur with similar frequency in both groups. Furthermore, non-specialists may interpret clinical findings differently and overtreat patients as having SSIs where specialists might consider an alternative diagnosis such as a suture granuloma, and therefore even if this information were available, it may not be accurate.

Coronavirus disease 2019 (COVID-19) has altered the public's perception of infection transmission and their behaviors related to this. All patients in the retrospective analysis group (receiving prophylactic antibiotic) were treated prior to the first major UK lockdown of the COVID-19 pandemic, and well prior to the time when widespread usage of face masks was adopted in the United Kingdom. In contrast to this group, most patients in the prospective “no chloramphenicol group” were treated during the pandemic, starting six months after the first UK lockdown, when routine eyelid operations were recommenced. This introduces several potential confounding factors which may influence any difference between the two groups. Patients in the prospective (no antibiotic) arm may have been washing their hands more and touching their face less in the wake of the pandemic, which could reduce the infection rates. On the other hand, mask wearing, particularly where ill-fitting, may redirect patients' breath into the eyes and actually increase the risk of infection after eyelid surgery. ¹² Likewise, contact between the top edge of the face mask and any lower lid wound or sutures could be potentially increase infection risk. We did not specifically give patients instructions about when to recommence mask wearing after surgery or gather data on this. Currently, there is no evidence by which to evaluate the potential impact of these issues on our SSI rates.

Healthcare practices have also changed because of the pandemic, with reduced face-to-face appointments; this may have altered surgical management. In our department we used predominantly braided absorbable sutures even before the pandemic, but further increased use of absorbable and rapidly absorbing sutures during the pandemic to prevent patients having to report purely for suture removal. All SSIs in our cohort occurred in patients with absorbable polyglactin sutures (Vicryl™ or Vicryl Rapide™, Ethicon, Bridgewater, NJ and Cincinatti, OH, USA), which are the sutures favored in our unit. These sutures were not routinely removed but were removed in some patients to manage infections suspected to be suture-related (7/17 patients). The majority of SSIs present three to 10 days post-operatively, and thus it is unlikely that routine removal of sutures would have occurred by this time.^4,13,14 However, because of the predominance of absorbable suture use in both study arms and overall low rate of infection in our cohort, we cannot draw conclusions about the association of different suture types on SSI rates from this study.

Conclusions

Post-operative wound infection rates in eyelid surgery are low. Routine use of prophylactic antibiotic agents in primarily closed surgical eyelid wounds does not adhere to international standards, and carries significant risks, including contributing to a world-wide antimicrobial resistance crisis. Early results suggest infection rates remain low without the use of prophylactic chloramphenicol in these patients, although numbers required to identify a statistical difference in rates are prohibitively high to be adequately assessed in a single unit. Multicenter collaboration may be beneficial in assessing this important issue to gain sufficient patient numbers to achieve statistical significance. Clinicians should consider withdrawing routine topical chloramphenicol prophylaxis from their routine practice to adhere to international guidelines and monitor their infection rates regularly.