Endophthalmitis: concepts in prevention

In this issue of Expert Review of Ophthalmology, Schwartz and colleagues review the current management of endophthalmitis and provide a useful clinical classification that is both authoritative and practical [1]. It is certain that the presentation of any patient with the signs and symptoms of endophthalmitis remains, to most ophthalmologists, a dreaded situation and, perhaps, even more so in the perioperative setting.

Perplexingly, many postoperative cases occur following apparently uncomplicated surgery without any obvious antecedent and, while some risk factors can usually be identified, it is often the same risk factors that are present in patients who experience an uneventful recovery [2].

The incidence of postoperative endophthalmitis appears to be procedure dependent, with the risk being greater in eyes that are undergoing therapeutic intravitreal injections (1%) [1,3], whereas endophthalmitis following vitrectomy occurs infrequently (0.04%) [4].

The risk of endophthalmitis following cataract surgery has been studied extensively, with many studies reporting an incidence of approximately one case in every 1000 surgeries [2]. However, even for this intraocular procedure, where multiple risk factors have been identified, there is considerable debate regarding the importance of which measures are most important in ameliorating the risk of infection [5]. There is a general agreement about irrigating the conjunctival sac with povidine iodine but, as to the merits of intracameral antibiotics, a split exists between Europe and North America [6,7].

Endophthalmitis following intraocular surgery, of course, should not be condoned but it is not unreasonable to anticipate that sporadic cases may occur. Management guidelines, as amplified by Schwartz and colleagues, should be directed to the acute event but, equally, strategies should be in place to detect and investigate all cases of postoperative endophthalmitis [1]. In most cases, this would require a dedicated infection control team to monitor and maintain protocols as part of a risk-containment policy. Implicit in this approach is an ability to identify cases of endophthalmitis that might occur infrequently as sporadic events, as well as those cases that might result from a system failure. The differentiation between a systems failure and an isolated event may be difficult or alarmingly obvious. Isolation of unusual organisms not normally regarded as skin commensals should alert the infection control team to the possibility of a breakdown in technique or sterility of the instruments or infusion fluid. Similarly, an outbreak or cluster of endophthalmitis cases that is associated with a particular procedure or surgeon may be an indication that a systemic failure exists.

Sparrow used a simulation model to determine the likelihood that clustering of endophthalmitis cases following cataract surgery is a random event [8]. For a set number of procedures and a set volume per surgeon, he determined the probability that an outbreak of endophthalmitis might occur relative to an expected annual frequency. These calculations established that for a single surgeon performing 1600 cataract surgeries, the chance of encountering four cases of endophthalmitis is 30%. Similarly, the chance of encountering two cases following 450 surgeries is 60%. However, in the scenario described by Sparrow, less than 1% of the surgeons would encounter five cases of endophthalmitis in every 427 surgeries. This approach can be usefully employed to establish guidelines for both the surgeon and facility regarding the actions that should be taken to manage a potential outbreak. Few would doubt the wisdom of canceling all surgery if confronted with an outbreak of five cases in 427 surgeries but, if lower rates of infection were documented, a strategy developed by Allardice and colleagues would seem to be sensible [9]. That group devised a ‘traffic light system’ to formalize a decision process regarding how and when to intervene. The basis for this model was derived from standard statistical theory, noting that if a random event occurs at a known frequency then the probability of observing a set number of cases is defined by the Poisson distribution. Thus, the decision to continue surgery would be ‘green’ if the observed frequency of postoperative endophthalmitis did not exceed the nominated background rate, ‘amber’ would reflect a decision to continue surgery but activate consultation and ‘red’ would halt surgery. The probability scores that were set by Allardice meant that if the probability was below 0.01 then surgery should stop, whereas amber was designated when the probability was between 0.01 and 0.05.

The intrinsic appeal of adopting a simplified process that alerts the clinician to potential risk does not, of course, abrogate the need to ensure that a robust infection control team exists within each and every surgical facility. The scope of such a team must include the development of infection control strategies, as well as ensuring that those strategies are adopted. If an infection is detected or some protocol breached, a comprehensive review should be performed.

Of course, not all patients with signs of inflammation in the postoperative period have endophthalmitis. Both toxic anterior segment syndrome and pseudoendophthalmitis following intravitreal injections can be confused with endophthalmitis and a strong case can be made to broaden the jurisdiction of the infection control team to include any patient with an adverse outcome. Such an approach achieves synergies in both reporting and monitoring, as well as fostering an ongoing review of critical control points.

The application of analyzing critical control points in the medical environment is relatively new but has been used successfully in other endeavours as a means to minimise adverse outcomes [10]. For example, this approach has been used by the National Aeronautics and Space Agency (NASA) to ensure that food provided for astronauts is not contaminated during space missions. In the ophthalmic setting, the utility of critical control points requires that each process in patient care is evaluated and, at those points of interaction, a standard of care is adopted, maintained and monitored. The value of this approach is that it is sufficiently flexible and applicable to modern ophthalmic practice. Furthermore, the outcome, that is reduced cases of postoperative infection, can be benchmarked.

It is probable that improvements in patient care are more likely to result from changes in quality assurance than the advent of new antibiotics.