Even though endophthalmitis is a rare complication of cataract surgery (recent studies report an incidence of 0.12% in the USA1 and 0.17% in Denmark2 ) its treatment continues to be a frustrating problem leading frequently to disastrous visual results. In this issue of the BJO (p 719) Okhraviet al assess the effect of a standard treatment protocol on the visual outcome of patients who have been diagnosed as having presumed bacterial endophthalmitis following intraocular surgery or penetrating ocular trauma. This protocol utilises early aggressive treatment with a standard regimen of high dose broad spectrum intraocular and systemic antibiotics. It also assesses the sensitivity of the organisms isolated to the specific antibiotics employed. In spite of aggressive and appropriate broad spectrum antibiotic therapy the final visual outcome was poor, with 55% of postsurgical and 40% of post-traumatic cases achieving a final visual acuity of 6/60 or less.

Why is the treatment of endophthalmitis still such an exercise in frustration? Are we somehow missing the point as clinicians, or is intraocular infection a problem in which a successful cure will always remain beyond our grasp? One difficulty is that endophthalmitis is not a single entity but is instead a heterogeneous group of infections with diverse epidemiological presentations and a varying spectrum of pathogenic organisms. In a review of 30 culture positive cases of endophthalmitis in Auckland from 1983 to 1991, Kent3 found 13 early onset postoperative, six delayed onset postoperative, five bleb associated, four post-traumatic, and two endogenous cases. To these five major categories should be added a sixth, infection by contiguous spread. In addition, bleb associated endophthalmitis can be divided into blebitis, which usually has a very favourable visual outcome, and bleb associated vitritis, which has a prognosis as guarded as early onset postoperative endophthalmitis. Endogenous endophthalmitis can be further subclassified into five categories as described by Greenwald et al.4 These are anterior focal, posterior focal, anterior diffuse, posterior diffuse, and panophthalmitis. All five categories have different clinical presentations, different visual prognoses, and very specific guidelines for diagnostic evaluation and antibiotic administration.

In addition to the clinical complexity of intraocular infection, every case of endophthalmitis presents the physician with three major issues which profoundly affect visual outcome but over which the clinician has no control. The first issue is the inherent pathogenicity of the infecting organism. Bacteria which are intrinsically highly pathogenic such as Pseudomonas aeruginosa and Streptococcus pyogenes produce severe intraocular infections that lead invariably to poor visual outcomes. Fortunately, the majority of cases of endophthalmitis are caused by Gram positive bacteria of relatively low pathogenicity. In the Endophthalmitis Vitrectomy Study5 Gram positive, coagulase negative staphylococci were the single most common group of isolates representing 70.0% of all positive cultures from endophthalmitis patients, 94.5% of whom were postoperative cataract surgery. Speaker et al 6 demonstrated by genetic analysis that 82.0% of coagulase negative staphylococci isolated from intraocular specimens in endophthalmitis are genetically indistinguishable from organisms on the patient’s lids. Since virtually all individuals haveStaphylococcus epidermidis colonising their eyelids, the source of the infecting organisms during cataract surgery is obvious. It must be emphatically stated, however, that other clinical forms of endophthalmitis are frequently caused by various other pathogens. Alfaro et al 7 reported 21 culture positive cases of post-traumatic endophthalmitis in which 25.6% of the isolates were Staphylococcus species, 20.6% wereStreptococcus species, and 14.7% wereBacillus species. And Thompson et al 8 isolated bacilli or staphylococci in 95% of culture positive cases of endophthalmitis following penetrating injuries with retained intraocular foreign bodies. Likewise, the overwhelming majority of cases of late onset postoperative endophthalmitis are caused by Propionibacterium acnes,9 while the pathogens responsible for bleb associated infections correspond with the organisms that produce acute purulent conjunctivitis—Haemophilus influenzae, Streptococcus pneumoniae, and Staphylococcus aureus. Endophthalmitis associated with infection by contiguous spread is most often caused by Pseudomonas species, while the bacteria responsible for endogenous endophthalmitis are a diverse group of organisms that include Bacillus species andKlebsiella pneumoniae as well as multiple streptococcal species and Staphylococcus aureus.10

The second major issue affecting visual outcome is the size of the inoculum of the infecting organism. Even a relatively benign pathogen such as Staphylococcus epidermidis can produce an endophthalmitis that is resistant to treatment if a sufficient number of organisms are introduced into the vitreous.11 All of the techniques for maintaining asepsis in the operating room are directed towards preventing such an event from occurring, but non-postoperative forms of endophthalmitis are exempt from these precautions. The third and perhaps most important issue affecting visual outcome is the time lapse between the onset of symptoms and clinical diagnoses and the initiation of aggressive appropriate therapy. Paradoxically, time is the only factor over which the clinician has a small measure of control. The faintest suspicion that the patient is developing endophthalmitis should lead to the immediate institution of a standardised diagnostic and treatment protocol such as the one described by Okhravi and colleagues in this issue.

Admittedly, there are still questions regarding the standard of therapeutic care. Are systemic antibiotics really necessary for the treatment of endophthalmitis? Has the question of when to do a vitrectomy been definitively answered? In spite of the results of the Endophthalmitis Vitrectomy Study12 these grey areas of clinical judgment continue to weigh heavily on clinicians. Perhaps it is the complex nature of endophthalmitis itself that prevents a standard protocol from being devised that can consistently produce a satisfactory visual outcome in this group of patients that are so difficult to treat. Unfortunately, even in the best of hands the results are still frustrating and the ideal treatment for endophthalmitis remains elusive.