- Top of page
- Patients and Methods
Purpose: To investigate the epidemiology of postoperative endophthalmitis (POE) following cataract surgery in Sweden during a 3-year period, using the framework of the Swedish National Cataract Register.
Methods: Clinically presumed cases of POE were reported in a prospective survey involving all Swedish ophthalmic surgical units except one. Data on results of the intraocular culture and visual outcome after infection, as well as patient age and gender and various elements of the surgical procedure, were collected.
Results: The nationwide incidence of POE was 0.0595%, representing 112 cases in 188 151 cataract operations. Gram-positive bacteria were the predominant aetiology, with an 84.6% share of culture-positive cases. A significantly decreased risk for POE was found for patients who had received prophylactic intracameral antibiotics (mainly cefuroxime) in comparison with those who had been treated with topical antibiotics only.
Conclusions: The prevalence of POE after cataract surgery in Sweden is at the lower end of the spectrum of incidence currently reported in the developed world. The administration of intracameral antibiotics may have contributed to these results.
Patients and Methods
- Top of page
- Patients and Methods
Since 1992, the NCR has collected data on cataract extractions performed in Sweden (Stenevi et al. 1995; Lundstrom et al. 2002). At present, all surgical units except one participate. Through comparison with another register source for intraocular procedures which is run by the Swedish Ophthalmological Society, the cataract operations reported to the NCR were found to cover 95% of the total number of cataract extractions in Sweden in the given 3-year study period. The surgical unit that did not participate in the study represented a further 1.9% of extractions. Since 1998, the participating clinics have been committed to supplying data on all POE cases, including culture-proven cases and presumed cases with a negative culture. For the latter entity no strict consensus regarding inclusion criteria exists in Sweden, but it is assumed that culture-negative patients with a slow or poor visual recovery are interpreted as POE cases (Montan et al. 1998). Reminders of the commitment to report POE cases are issued regularly. The case report consists of two forms, the first of which requires the date of suspected diagnosis and intraocular culture samples (whether positive, negative or not taken), while the second, to be completed about 3 months later, asks for information on the type of isolated organism, the prophylactic regime and present visual acuity. At the end of every year, all clinics are asked to verify the number of cases of POE, including those clinics that have not reported any cases primarily. Using a unique identification number related to the consecutive order of the operation at the respective clinic, the infected cases can be traced in the NCR, thereby allowing for analyses of data on operation technique, patient gender and age. There is no predefined maximum delay between the operation and the onset of symptoms. With the closing of the database in April 2003, the range of postoperative follow-up time for cases operated within the given timeframe is roughly 1–4 years.
One novel aspect of the 1998 survey (Montan et al. 2002a) required each clinic to report its current use of prophylactic method in the month of December and to state whether this practice had been subject to change in the course of the preceding year. Disinfecting procedures vary a lot between different units, regardless of other prophylactic measures. Povidone iodine is not commonly used, as it is not commercially available in Sweden. (To the best of our knowledge it was used in only one unit during the study period.) To avoid the risk of comparing too many different protocols, the material was dichotomized into intracameral antibiotics administered as bolus injections or in infusion fluid (regardless of additional topical anti-infectives) and ocular surface antibiotics only, given either as topical antibiotics or as subconjunctival injections, or both.
Guidelines concerning diagnostic and therapeutic management of POE in line with the evidence presented by the Endophthalmitis Vitrectomy Study Group (1995) and Barza et al. (1997) are accessible to the ophthalmic community in an extensive document that has been endorsed by and published on the home page of the Swedish Ophthalmological Society since 1996 (http://www.swedeye.org). Gold standard culturing methods including solid media and enrichment broths for both aerobic and anaerobic cultures of intraocular samples from endophthalmitis cases were described in a document published in 1994 by the Swedish Institute for Infectious Disease Control and are intended to serve as guidelines for all microbiological units in Sweden. Data for the registers were collected after the approval of the local ethical committee had been gained.
All statistical evaluations were performed using spss Version 10.0. Pearson chi-squared test was employed in univariate analyses to explore the association between categoric variables and the development of POE, and confidence intervals for odds ratios were calculated. A logistic regression model was subsequently used. A p-value of less than 0.05 was defined as significant.
- Top of page
- Patients and Methods
In 1999–2001, 188 151 cataract operations and 112 cases of POE were reported to the NCR, reflecting an overall incidence of POE of 0.0595% of cataract operations. Community-run surgical units reported 153 041 operations and 91 cases of POE (0.0595%), while private surgical units reported 35 110 operations and 21 cases of POE (0.0598%). Three of the 112 cases were lost to follow-up, meaning that 109 cases only were eligible for analyses relating to POE data and risk factor evaluation. The median latency between the operation and the diagnostic procedure was 6 days. A late diagnosis, beyond 6 weeks after surgery, was identified in six cases.
Gram-positive bacteria were the dominating aetiology, comprising 84.6% of culture-proven cases. The share of enterococci among causative bacteria was 25.3% (Table 1).
Table 1. Aetiology of postoperative endo-phthalmitis.
|Type of organism||Total number|
| Coagulase negative staphylococci (CNS)||30|
| Staphylococcus aureus||6|
| Other streptococci*||14|
| P. acnes||3|
| Turicibacter sanguinis||1|
| Pseudomonas sp.||4|
| H. influenzae||4|
| Serratia marscesens||2|
| E. coli||1|
| Proteus sp.||1|
| Morganella sp.||1|
| No growth||14|
| No culture taken||3|
| Missing data||4|
In 99 of 112 cases, the visual outcome was reported to the database. A 20/40 acuity or better was reached by 28%. Roughly half of the study group, 52.5%, had a final outcome of less than 20/200, implying that the overall risk of sustaining severe visual impairment due to POE was three in 10 000 cataract procedures.
The inquiry data on the use of prophylactic antibiotics were analysed only from those clinics where the same regime had been in use during the entire previous year. The frequency of POE in patients who had received intracameral antibiotics was significantly lower (81 cases in 151 874 procedures, 0.053%) than it was in patients who had not been given intracameral antibiotics (15 in 6805 operations, 0.22%, p < 0.001) (Table 2). This difference was also found to be statistically significant in a logistic regression analysis (p < 0.001). No other significant risk factors among the tested variables of age, gender or type of surgery were identified in the logistic regression model (Table 3). There was no significantly increased risk associated with any of the different lens materials (Table 4).
Table 2. Use of prophylactic antibiotics in clinics with unchanged regimes during an entire year.*
| ||POE cases/ total number||(%)||Odds ratio (95% CI)|| ||p-value|
|Intra-cameral antibiotics†||81/151,874||(0.053)||1.0|| || |
|No intracameral antibiotics‡||15/6805||(0.220)||3.649||(2.291–5.812)||< 0.001|
Table 3. Patient history, operation technique and association with POE.
|History||POE cases/ total number||(%)||Odds ratio (95% CI)|| ||p-value|
| 0–84 years||92/155,175||(0.059)||0.996||(0.914–1.086)||0.926|
| 85 years +||20/32 972||(0.061)||1.0|| || |
| Females||74/124 381||(0.059)||1.001||(0.773–1.297)||0.992|
| Males||38/63 744||(0.060)||1.0|| || |
|Type of surgery|
| Phaco + PC IOL||103/182 064||(0.056)||1.0|| || |
| Other than phaco*||6/6069||(0.099)||1.707||(0.784–3.717)||0.178|
Table 4. IOL materials and frequency of POE.
|IOL material||POE cases/ total number||(%)|
| HSM PMMA||7/6059||(0.116)|
| Acrylic||80/142 574||(0.056)|
| Silicone||12/21 225||(0.056)|
- Top of page
- Patients and Methods
This is the largest prospective epidemiological study on POE following cataract surgery to be presented in the last decade. Although there may be some under-reporting of both cases and procedures in undertakings of this kind, we believe that the data quality of the current work is high. Consistent and continuous reporting by Swedish ophthalmological surgical units to the NCR for almost 10 years substantiates this assumption. Moreover, the practice of sending repeated reminders to participants and making a yearly request for verification of submitted data from all clinics most likely contributes to a high level of validity in the reports. As for the enquiry on prophylactic regimes, it should be emphasized that the fact that clinic policies in this respect must be accepted by every surgeon concerned lends a high level of credence to these data as well. In addition, the choice of POE criteria that included both culture-negative and late-onset cases prevented any premeditated exclusion of cases and the incidence found should be afforded a high degree of credibility.
As there was no collection of data on some parameters that may influence the frequency of POE, such as presence of diabetes or perioperative complications (Kattan et al. 1991; Menikoff et al. 1991), the results of our risk factor screening must be interpreted with much caution. Nevertheless, we believe there is an actual difference in prevention effectiveness between intracameral antibiotics and ocular surface antibiotics, for a number of reasons. First, the present results were generated in a project with a prospective design, but naturally our data cannot present proof of concept because they are not derived from a true treatment trial with balanced randomization between strict prophylactic protocols. Second, the frequencies of POE in the respective treatment groups were found to be almost identical to those reported in large single-centre studies from Sweden, where either subconjunctival, topical anti-infectives or the combination of a preoperative disinfectant and intracameral cefuroxime were used (Montan et al. 1998, 2000, 2002b). Furthermore, retrospective data from a large scale, multicentre German investigation point to an advantage with intraocular over topical antibiotics as a preventive measure (Schmitz et al. 1999). Third, the apparent changes in aetiology, seemingly harmonizing with the targeted bacterial spectrum of cefuroxime, indicate that this mode of prophylaxis exhibited an impact on POE morbidity in the study period. When comparing the causes of POE in the current study with those in the 1998 survey (Montan et al. 2002a), shifts away from infections in the total cataract population due to coagulase negative staphylococci (CNS) and S. aureus were noted, but none of the differences were statistically significant (p = 0.08 and 0.19, respectively, data not shown).
Conversely, the share of enterococci, a cefuroxime-resistant species, among causative bacteria in culture-proven cases grew from 12.2% in the 1998 report (Montan et al. 2002a) (data not shown) to 25.3% in the present paper. However, the frequency of POE due to enterococci did not change in the general cataract population (i.e. 5/54 666, 0.009% versus 23/188 151, 0.012%). These shifts in aetiology are believed to have influenced visual outcomes following POE incidents in the present investigation, as only 28% of patients reached 20/40, compared to roughly 50% in the multicentre project conducted by the Endophthalmitis Vitrectomy Study (EVS) Group (1995). This is most likely explained by the difference in proportions of culture-negative cases and cases infected by CNS, where subjects almost invariably recover good visual acuity (Endophthalmitis Vitrectomy Study (EVS) Group 1995; Montan et al. 1998, 2000, 2002a, 2002b). Such patients accounted for 80% of cases in the EVS, while they make up only 40% in the present material.
In our previous research, a significantly higher incidence of POE was found with hydrogel and polymethylmethacrylate (PMMA) than with acrylic intraocular lenses (IOLs) (Montan et al. 2002a), but this was not confirmed in the present study. In the current study there was a somewhat higher frequency of POE found with heparin surface modified (HSM) PMMA IOLs and among patients operated with techniques other than phacoemulsification plus posterior chamber (PC) IOL. However, these groups were small, differences were not statistically significant and procedures where HSM PMMA IOLs and techniques other than phacoemulsification were used are likely to be associated with other risk factors, such as capsular breaks, which were not investigated parameters in the present study. Recent evidence suggests that an injected IOL may decrease the risk for POE, whereas a clear corneal incision and, especially, a temporal corneal incision may in fact increase it (Schmitz et al. 1999; Cooper et al. 2003; Mayer et al. 2003; Nagaki et al. 2003). It is thus clear that a substantial number of operating technique variables need to be evaluated in a multivariate analysis before the actual role of various IOL materials in POE development can be defined. With the aim of gaining further knowledge in this respect, registration on the NCR since 2002 has incorporated additional variables, such as incision site, the use of injectors and the presence of intraoperative complications. We assume that after a 3-year period enough data will have accumulated to allow for analyses of these issues in relation to the development of POE.