Estimating the Incidence of Endophthalmitis Using a National Medical Database

The study by Cedrone et al. in this issue reports the annual incidence of endophthalmitis in Italy over a five-year time period (1999–2003). In addition to addressing this primary aim, this manuscript is informative for two reasons. First, it highlights a substantial increase in the annual number of endophthalmitis cases nationally, and second, it provides a well-executed example of how national administrative databases can be used as powerful research tools, even though they are designed for other reasons.

The Cedrone study reports a 30% increase in the annual number of endophthalmitis cases during a five-year period. It also reports the annual frequency of various “high risk” events including corneal ulcers and open wounds of the eye, and four sets of surgical procedures for which endophthalmitis may be a post-surgical complication. The authors conclude that while the total number of endophthalmitis cases in the country increased between 1999–2003, the frequency of events where the risk of endophthalmitis is “high” increased at a similar rate. Specifically, the frequency of cataract surgery increased 33% during the same time period. Recognizing that in the Cedrone study a causal link cannot be drawn between the increase in cataract surgeries and the increase in the total number of endophthalmitis cases, it is reasonable to assume that much of the increased number of endophthalmitis cases in this study did, in fact, result from the increased number of cataract surgeries. One would expect a similar rise in the number of endophthalmitis cases in any country where the cataract surgical volume increases substantially. Cedrone et al. note that while the cataract surgery rate (CSR) in Italy increased substantially during this time period, it remains below the CSR for some other European countries such as Sweden. Thus, the CSR in Italy may continue to rise.

Significant emphasis has been placed on increasing the CSR in many countries, particularly in developing countries where the surgical backlog is quite large. In recent years, the reported rate of cataract surgery has increased in some developing countries, such as India, as access to surgery has improved. With the current emphasis on improving access to surgery, it is likely that the annual number of cataract surgeries worldwide will rise dramatically. Furthermore, it is projected that over the next 25 years the U.S. population aged 65 and older will almost double, and the population aged 85 and older will almost quadruple. This aging of our population will likely result in a proportional increase in the annual volume of cataract surgery.

Even in the absence of an increase in the rate of endophthalmitis following cataract surgery, the substantial rise in the number of cases presumably linked to cataract surgery highlights the importance of finding measures to reduce the risk of post-cataract surgery endophthalmitis. An endophthalmitis prevention trial conducted by the European Society for Cataract and Refractive Surgery (ESCRS) reported that the use of intracameral cefuroxime at the close of cataract surgery reduced the rate of post-cataract surgery endophthalmitis compared to no peri-operative antibiotic use.¹ The endophthalmitis rate was approximately 1/1400 among surgeries where intracameral cefuroxime was used without peri-operative levofloxacin drops and approximately 1/2000 among patients receiving intracameral cefuroxime and peri-operative levofloxacin drops, compared to 1/300 in patients not receiving intracameral antibiotics. Even lower rates of endophthalmitis have been reported in Sweden, where intracameral cefuroxime administration has become routine in cataract surgery.² Despite these findings, few cataract surgeons in the United States have adopted this strategy. The American Society of Cataract and Refractive Surgeons (ASCRS) recently surveyed its members and found that of 1312 respondents, only 16% of surgeons utilized any direct intracameral antibiotic more than one year following the ESCRS study results.³

Several factors may explain this lack of uptake. First, in the ESCRS study the remarkably high rate of endophthalmitis in the control group (about 1 per 300), causes some to believe that the findings from the European study are not generalizable. Second, there appears to be an unwillingness among American surgeons to take on the responsibility of compounding the antibiotic for administration. Although cefuroxime has a favorable safety profile, surgeons express concern that the rate of compounding errors leading to an adverse event could outweigh the risk of endophthalmitis. Third, there is concern about the use of a cephalosporin in patients who might have a penicillin allergy. These last two concerns are supported by the ASCRS survey results, where 45% of individuals not using intracameral antibiotics indicated the reason was concern about risk.³ It is likely that uptake of cefuroxime will be limited in developing country settings as well, since access to pharmacies who will perform compounding often is limited, particularly in outreach (surgical camp) settings.

Three recently published studies illustrate that some surgeons are investigating other antibiotics for intracameral use during surgery. Each of these studies investigated the safety of intracameral injection of moxifloxacin ophthalmic solution, either diluted to 100 ug/0.1 mL or withdrawn directly from the bottle,⁴, ⁵, ⁶ and all reported no safety issues associated with intracameral injection. Endophthalmitis prophylaxis during cataract surgery is an important issue to all cataract surgeons, and the current study provides yet more evidence that new prophylactic measures for endophthalmitis will become increasingly important as the annual number of cataract surgeries worldwide continues to rise.

This study also provides a useful example of how national medical databases can be exploited to investigate basic research questions, despite the limited search capabilities of the database. The Italian database (Scheda di Dimissione Ospedaliera (SDO)) does not allow for cross-linked queries. For example, one could not query the database to determine the number of patients who had cataract surgery and developed endophthalmitis. Interpretations must be made by comparing changes in frequency of two specific factors (in this cases, frequency of endophthalmitis and cataract surgery), separately. Such databases cannot be used to investigate outcomes that may be achieved from numerous pathways. For instance, one could not use queries of the annual number of open-heart surgeries and the annual number of in-hospital deaths and to drawn any plausible conclusions. National databases are useful for investigating changes in frequencies of specific events over time (either procedures or outcomes) and can be used as a first level of investigation into whether associations may exist between changes in procedure rates, and specific, well-defined outcomes with a limited number of causal pathways.

Of course, this methodology is not without limitations. Many of these limitations have already been noted by Cedrone et al. Perhaps the most important limitation is recognizing that causal associations cannot be evaluated, regardless of the sensitivity and specificity of the presumed associations between the procedure and outcome being examined. For example, in the current study, no conclusions can be drawn regarding the reason for the increased risk of endophthalmitis among males. The authors speculate on a number of factors, including higher prevalence of open wounds among men and increased frequency of “high risk” procedures, aside from cataract surgery, among men. However, the limitations of the database query capabilities prevent drawing even loose connections between these varied factors.

In summary, the Cedrone study uses a thoughtful approach for investigating the possibility of an increase in the rate of endophthalmitis and reports a substantial increase in the annual number of endophthalmitis cases. This manuscript is also a useful case-study in the potential use of national databases as a starting point for addressing important, practice-related research questions.