Significant Nonsurgical Risk Factors for Endophthalmitis after Cataract Surgery: EPSWA Fourth Report

Abstract

purpose. Endophthalmitis remains a devastating complication of cataract surgery, despite improved methods of prophylaxis and surgical technique. The current study was conducted to identify sociodemographic, environmental, and clinical risk factors for the development of postoperative endophthalmitis, using population-based administrative data from Western Australia.

methods. The Western Australian Data Linkage System identified all patients who underwent cataract surgery, along with those in whom postoperative endophthalmitis subsequently developed, from 1980 to 2000 inclusive. Cases of endophthalmitis were cross-referenced with other sources and validated by medical record review. After selection and preliminary analysis of potential risk factors, multivariate logistic regression modeling was used to estimate odds ratios for the selected variables.

results. Over the 21 years, 210 cases of endophthalmitis occurred after 117,083 cataract procedures, yielding a cumulative incidence rate of 1.79 per 1000 procedures. The incidence of endophthalmitis decreased for extracapsular extraction over the whole period, but not for phacoemulsification over the recent 12 years. There was no risk-adjusted difference in the incidence rate of endophthalmitis for the various cataract surgery procedure types. However, a significantly higher risk was found in patients aged over 80 years, in having surgery in private hospitals, and to a lesser degree in having same-day surgery and surgery in winter. Cataract surgery with lacrimal or eyelid procedures dramatically increased the risk of endophthalmitis.

conclusions. It may be possible to reduce the incidence rate of postoperative endophthalmitis by almost 80% with a systematic approach to the management of elderly patients, hospital stay, and clinical protocols.

Postoperative endophthalmitis remains a devastating outcome after cataract surgery, despite improved methods of prophylaxis, surgical technique, and treatment. This complication causes significant morbidity and distress and often severe visual impairment or blindness. ^{1 2 3 4 5}  

The reported incidence of endophthalmitis after cataract surgery varies considerably around the world 1 . ^{2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28} Occurring about once every 500 operations, the relatively low incidence of endophthalmitis makes it difficult to identify the risk factors clearly. ^{3 29} Previous studies of risk factors were limited by using only data from individual hospitals and centers or groups of surgeons. ^{30 31 32 33} This results in a low number of cases and reduced external validity, making it difficult to compare results. 

Our purpose was to identify and quantify a range of risk factors for the development of postoperative endophthalmitis using the Western Australian Data Linkage System as part of the Western Australian Safety and Quality of Surgical Care Project. ^{29 34 35} The significant advantage of our Data Linkage System is that it completely encompasses a geographically isolated and generally stable population of 1.9 million people, providing reliable denominators for our statistical model. 

Previous studies have focused primarily on surgical variables as risk factors for postoperative endophthalmitis. ^{30 31 32 33} Several studies have reported a significant reduction in the incidence of endophthalmitis after the transition from intracapsular to extracapsular extraction. ^{3 9 11 29} However, despite the introduction of the phacoemulsification technique of cataract surgery, heralding the advent of small incision, sutureless surgery, and the widespread adoption of same-day surgery, it remains controversial whether these changes have had any impact on the rate of endophthalmitis. ^{3 5} Also, the potential influence of social, demographic, and environmental risk factors have received little attention to date. Because these uncharted risk factors may confound the effect of surgical technique on endophthalmitis, the overall importance of surgical factors remains unknown. 

This study addresses the paucity of knowledge about sociodemographic, environmental factors (such as the timing and location of surgery), and comorbidities that may contribute to the development of endophthalmitis. 

Methods

Ethics Approval

Approvals for accessing patient health records were obtained from the University of Western Australia Ethics Committee, the Confidentiality of Health Information Committee of the Western Australian Department of Health, and individual hospitals. The research was conducted in accordance with the Declaration of Helsinki. 

Data Selection

All cataract operations in Western Australia, including same-day surgery, can be performed only in licensed facilities. These facilities, or hospitals, are required by law to submit morbidity data for each patient discharge, and this information provides the core data set of the Western Australian Data Linkage System. Using record linkage, data from all patient discharges enabled us to identify all patients who underwent cataract surgery in Western Australia from 1980 to 2000 inclusive. Records were selected using the International Classification for Diseases (ICD) procedure codes for cataract surgery: ICD-9 and ICPM for 1980 to 1987; its modification, ICD-9-CM for 1988 to 1998; and ICD-10-AM for 1999 forward (Appendix A). ^{36 37 38 39 40} The data set consisted of all hospital admissions for patients who had cataract surgery during the study period. The ICD diagnosis codes for endophthalmitis were used to identify patients with an admission for endophthalmitis after cataract surgery as summarized in Appendix A. 

Data Validation

All cases of postoperative endophthalmitis cases were cross-referenced with specific clinical databases located in the Departments of Microbiology and Anesthetics at Royal Perth Hospital (Western Australia, Australia). Two vitreoretinal subspecialists managed almost all cases of endophthalmitis at some point. To ensure further complete case identification we reviewed their surgical logbooks over the 21-year period. All potential endophthalmitis cases were then validated by medical record review. ⁴¹ Three cases of endophthalmitis were excluded from the analysis because of inadequate clinical data. 

Selection of the Variables for Statistical Analysis

Age was stratified into two groups: younger than 80 years, or 80 years old or older. Preliminary analysis indicated that age as a continuous variable, or divided into four groups (0–49, 50–59, 60–69, and 70–79 years) had no significant effect on the risk of development of postoperative endophthalmitis. 

Patients were classified as residing in metropolitan or country areas (rural and remote) according to the postcode of their residential address in the Health Zone classification system of the Western Australian Department of Health. Although there were 359 procedures (0.3% of 117,083) for which we had no postcode information to determine residential area, none were cases of endophthalmitis. 

Health insurance type was coded into three categories: public (public patients), private (privately insured patients), and other. Patients covered by workers’ compensation, motor vehicle injuries, or veteran (returned military service personnel) insurance schemes were classified as private because these patients were treated privately. 

The Dartmouth-Manitoba adaptation of the Charlson Index of comorbidity was used to measure patient comorbidity (Appendix A). ⁴² The presence of comorbidity was defined by the coding of any of the 17 conditions from the Charlson Index at either the index admission for cataract surgery, or any hospital admission within the previous year. Comorbidity was included in the final model as a dichotomous variable after preliminary analysis had shown that the results remained unchanged, regardless of whether comorbidity was analyzed as a continuous or dichotomous variable. Record linkage to the Western Australian Cancer Registry also allowed identification of patients with a cancer diagnosis before or during the cataract surgery admission. Cancer diagnosis was also used as a dichotomous indicator. 

We differentiated hospital of surgery into private versus public and rural versus metropolitan. Public hospitals were subdivided into metropolitan public, rural public, and teaching public hospitals. Hospital of surgery was also differentiated according to the cumulative number of cataract operations performed over the 21-year period. One hospital had performed 21% of all cataract procedures and was classified as the largest hospital. The next group of hospitals was classified as large with 2000 or more procedures, medium with 1000 or more, and small hospitals with less than 1000. Fourteen hospitals were excluded from the analysis because they had each coded less than 10 cataract procedures and because of the nature of the hospitals (for example, providing only maternity or psychiatric care) we suspected coding errors. No cases of postoperative endophthalmitis originated from these hospitals. 

Length of hospital stay for cataract surgery was categorized into five groups (same day and 1, 2–5, 6–7, and 8 or more days) after preliminary analysis showed that a continuous variable or other groupings had no effect on the outcome. 

Surgical technique was differentiated into three categories—intracapsular, extracapsular/phacoemulsification, and other cataract extraction/lens-related procedures—because analysis revealed no significant difference in the mean incidence rate of endophthalmitis between phacoemulsification and extracapsular procedures, or between other cataract extraction and lens-related procedures. 

Timing of surgery was analyzed according to the year and season of the admission date for surgery. Seasons were defined according to those in the Southern Hemisphere: winter, June to August; spring, September to November; summer, December to February; and autumn, March to May. 

Statistical Analysis

Patients who underwent cataract surgery and had postoperative endophthalmitis were classified as cases, and those who underwent surgery but did not have endophthalmitis were others. Analyses were performed on computer (SPSS; SPSS Science, Chicago, IL; and Stata; Stata Corporation, College Station, TX). ^{43 44} A multivariate logistic regression model (referred to as the model) was used to estimate odds ratios (OR) and 95% confidence intervals (CI) for each risk factor. Analysis was based on the procedure as the unit of analysis, with adjustment for possible clustering of multiple procedures in a patient. The clustering adjustment used generalized estimating equations, which relaxed the independence assumption and required only that the procedures be independent across patients (the clusters). This produced correct standard errors, probabilities, and confidence intervals, leaving the odds ratios and beta coefficients unchanged. ⁴⁵  

Results

There were 117,083 cataract procedures performed in 76,424 patients in Western Australia from 1980 to 2000 inclusive. There were 210 cases (occurring in 209 patients) of endophthalmitis after cataract surgery. The cumulative incidence rate of postoperative endophthalmitis was 1.79 per 1000 procedures over the 21-year period. 2 shows characteristics of the two groups of patients and the results of the multivariate logistic regression analysis. 

Patient Characteristics

Cataract surgery was more commonly performed among female patients and patients with private health insurance. Older age (80 years and older) was an important risk factor for developing endophthalmitis after cataract surgery, independent of other risk factors, including comorbidity (OR 1.50; 95% CI 1.13–1.99). No significant association was found between gender, comorbidity, health insurance type, or the locality of the patient’s residence and the risk of endophthalmitis. 

Hospital of Cataract Surgery

More cataract procedures were performed in private hospitals and metropolitan hospitals than in public hospitals and hospitals in rural areas. Large hospitals performed two thirds of all cataract surgeries. 

Procedures performed in private hospitals had a significantly higher risk of postoperative infection than did those performed in public hospitals (OR 2.38; 95% CI 1.32–4.27). However, hospital location, the type of public hospital (metropolitan versus teaching) or the volume of cataract surgery performed at each hospital had no effect on the incidence rate of endophthalmitis. 

Same-day cataract surgery was performed in 43% of the cases and 40% of the others. Most inpatient procedures had a length of stay of 1 to 3 days (2 , 1 ). Despite a steady decline in the length of stay since 1980, 22% of cataract operations in 2000 were still performed with inpatient admission, with an average length of stay of 1 day 1 . 

Patients undergoing same-day cataract surgery were more prone to development of endophthalmitis than those with inpatient stays of 2 to 5 days, but over the 21 years of data the difference did not achieve statistical significance (OR 1.42, 95% CI 0.95–2.13, P = 0.09). However, for the period 1990 to 2000, the risk associated with same-day surgery was statistically significant (OR 1.62, 95% CI 1.01–2.61, P = 0.047). The difference between the entire 21 years’ and the latter 10-years’ results was due to the virtual nonexistence of same-day surgical procedures before 1990 1 . 

We found no other interactions between year of surgery and length of hospital stay. There was no trend effect of the length of hospital stay on the outcome. The higher risk observed with a prolonged hospital stay of eight or more days represents delayed discharge due to the development and treatment of endophthalmitis during the same admission as that for cataract surgery. 

Surgical Characteristics

There was a decline in the incidence rate of endophthalmitis after extracapsular surgery in Western Australia over the 21-year period 2 , with a five-yearly OR of 0.75 (95% CI 0.57–0.98, P = 0.04) after adjustment for patient characteristics and hospital of surgery. No decline in the incidence rate was observed during the 12 years of phacoemulsification surgery. 

The incidence of postoperative endophthalmitis decreased by 50% during the transition from intracapsular to extracapsular extraction methods (2.92 vs. 1.46 cases per 1000 procedures). In contrast, there was no reduction in the incidence of postoperative endophthalmitis from the extracapsular to the phacoemulsification technique (OR 0.80, 95% CI 0.59–1.09). The intracapsular procedure and the combined group of other cataract extraction and lens-related procedures had a higher risk for endophthalmitis than did extracapsular and phacoemulsification surgery, independent of year of surgery, demographic variables, and hospital of surgery. However, after adjustment for all the other risk factors in the model, this association was no longer significant 3 . 

Approximately 10% of patients in endophthalmitis cases and 5% of those in others had undergone both cataract surgeryand another eye procedure during the same admission. Patients who had lacrimal or eyelid procedures during the same hospital admission were at a significantly higher risk of development of endophthalmitis than those who had cataract surgery only (OR 23.50, 95% CI 8.50–64.98). Although patients undergoing a vitreoretinal procedure during the same admission appear to be at risk, review of those medical records revealed that many of the vitreoretinal procedures were undertaken as treatment for endophthalmitis that developed during the same admission as that for cataract surgery. 

Environmental Factors

Cataract procedures were fairly evenly distributed across seasons, with summer being the least likely season for patients to undergo cataract extraction. However, more endophthalmitis cases resulted from surgery performed in winter and autumn than from those performed in spring and summer. Cataract surgeries performed during winter had a higher risk of endophthalmitis than those performed during spring (OR 1.48, 95% CI 1.00–2.18). 

Discussion

These data confirm our previous finding that a “center effect” has a significant bearing on the incidence of endophthalmitis after cataract surgery. ²⁹ There was also a large difference in the risk between public and private hospitals, independent of the other factors. Differences in the systems of patient care may be the underlying issue, particularly as we also found a lower risk for inpatients in the 1990 to 2000 period. This suggests that the perioperative and early postoperative care of patients may be an important factor in the risk of endophthalmitis, and this may be related to the reliability of administration of prophylactic antibiotics. 

It is not feasible to control all the identified significant risk factors; however, some of those described are manageable. It is ill advised to perform lacrimal or eyelid procedures at the time of cataract surgery, but it may be useful to increase the length of hospital stay from same day to at least 2 days, and to optimize perioperative clinical protocols, especially in private hospitals. An important finding was that patients aged 80 years or more were more likely to develop postoperative endophthalmitis, independent of comorbidity, and other factors. This suggests that additional measures could be beneficially directed toward this age group. 

Our findings suggest that to minimize the risk of endophthalmitis, the over 80-year age group would benefit from avoiding winter surgery and being admitted to hospital for a few days for cataract surgery. The model allows us to estimate risks under various scenarios for this age group, as demonstrated in 4 . Estimates from the model suggest a possible reduction in the endophthalmitis incidence rate of 23% to 78% for the 80-year or older age group, or 68% to 78% overall. In 2000 the 80-year or older group had 26.5% of the 11,439 cataract operations, of which 24.5% were inpatients (33% of the private patients, 14% of the public patients). If the remaining 75.5% elderly patients were also operated on as inpatients, an additional 2300 inpatients would double the number of inpatient admissions overall, adding approximately AUD$2 million to the health expenditure in Western Australia each year. Alternatively, elderly patients may benefit from more specific additional chemoprophylactic measures before, during, and after cataract surgery. Our findings also suggest that there is potential to greatly reduce the risk of postoperative endophthalmitis by reducing the differences in the hospital systems between the public and private sectors. Our case–control study and ongoing research should enhance our understanding of some of the factors responsible for the hospital system effect. 

Consistent with previous findings, our study has shown that the incidence of postoperative endophthalmitis decreased by half during the transition from intracapsular to extracapsular extraction methods. ^{3 9 11 29} However, no further reduction in incidence from extracapsular to phacoemulsification extraction was observed 2 . The significant decrease in endophthalmitis after extracapsular extraction occurred over a long period, suggesting that a refinement of systems and surgical technique have occurred gradually over time. Given that the reduction in the incidence of endophthalmitis after extracapsular surgery took 21 years (with no significant reduction during the first 10 years), the 12 years in which phacoemulsification was performed may be too short for any evident reduction in endophthalmitis incidence to be observed. 

The overall risk of postoperative endophthalmitis was higher after intracapsular extraction, as well as other cataract and lens-related procedures, compared with extracapsular and phacoemulsification procedures. However, other factors such as the hospital of surgery, the length of stay, and undergoing other eye procedures appear to be more predominant risks than the actual type of cataract surgery. These findings emphasize the importance of adjustment for other risk factors when examining the impact of changing surgical practice on outcome. 

The finding that season was a risk factor is consistent with existing literature on seasonal patterns of morbidity and mortality. Hippocrates observed over 2000 years ago that “spring is the most healthy, and least mortal.” Research has shown seasonal variation in the occurrence of infectious diseases, with the highest incidence during winter. ⁴⁶ These patterns may reflect greater exposure to infectious diseases in winter and human responsiveness to changes in photoperiod, temperature, and rainfall. ⁴⁶ Although the seasons were a significant risk factor, they did not account for the temporal fluctuation in endophthalmitis we have described elsewhere. ²⁹ Season may be a de facto marker of other environmental indices, and therefore further study is required to clarify this association between endophthalmitis and weather. 

Our findings suggest that simultaneous cataract and lacrimal or eyelid surgery is ill advised. They also suggest that the incidence of postoperative endophthalmitis may be reduced with a systematic approach to elderly patients, the length of hospital stay, and relevant clinical protocols. 

Appendix A

Cataract procedures included intracapsular extraction of lens (ICD-9 codes: 5-144; ICD-9-CM codes: 13.11, 13.19; ICD-10-AM codes: 42698-00, 42702-00, 42702-01); extracapsular extraction (linear/aspiration/other) (ICD-9 codes: 5-142, 5-145; ICD-9-CM codes: 13.2, 13.3, 13.51, 13.59; ICD-10-AM codes: 42698-01, 42702-02, 42702-03, 42698-04, 42702-08, 42702-09); phacoemulsification (ICD-9-CM codes: 13.41-13.43; ICD-10-AM codes: 42698-02, 42702-04, 42702-05, 42698-03, 42702-06, 42702-07); other cataract extraction (ICD-9 codes: 5-146; ICD-9-CM codes: 13.64, 13.65, 13.66, 13.69; ICD-10-AM codes: 42731-01, 42698-05, 42702-01, 4270210, 42702-11); and lens-related operations (ICD-9 codes: 5-147, 5-149; ICD-9-CM codes: 13.70-13.72, 13.9; ICD-10-AM codes under categories of 193, 194, 201-203). Endophthalmitis cases were identified using ICD-9 diagnosis codes: 360.0, 360.1; ICD-9-CM codes: 360.00-360.04, 360.11-360.19; and ICD-10-AM codes: H44.0, H44.1. 

The 17 categories of diseases included in the Charlson Index are: myocardial infarction; congestive cardiac failure; peripheral vascular disease; cerebrovascular disease; dementia; chronic pulmonary disease; rheumatological disease; peptic ulcer disease; mild liver disease; diabetes (mild to moderate); diabetes with chronic complications; hemiplegia or paraplegia; renal disease; any malignancy, including lymphoma and leukemia; moderate or severe liver disease; metastatic solid tumor; and acquired immunodeficiency syndrome. 

Appendix B

Team EPSWA: Other Members

Ian Constable (Sir Charles Gairdner Hospital; St. John of God Hospital, Subiaco; Lions Eye Institute; The University of Western Australia), Ian McAllister (Royal Perth Hospital; St. John of God Hospital, Subiaco; Lions Eye Institute; The University of Western Australia), Chris Kennedy (Fremantle Hospital; St. John of God Hospital, Subiaco), Tim Isaacs (Royal Perth Hospital; St. John of God Hospital, Subiaco; Lions Eye Institute; The University of Western Australia), John Pearman (Royal Perth Hospital), D’Arcy Holman (School of Population Health; The University of Western Australia). 

Other Contributors to the Study

Bridget Mullholland (Royal Perth Hospital), Max Bulsara (School of Population Health; The University of Western Australia), Arem Gavin (School of Population Health; The University of Western Australia), Ferenz Kosaras (Royal Perth Hospital), and The Health Information Centre of the Western Australian Department of Health. 

³ Additional members of the Endophthalmitis Population Study of Western Australia (EPSWA) team are listed in Appendix B.

 Supported by Australian National Health and Medical Research Council Project 110250.

 Submitted for publication September 9, 2003; revised January 4 and 11, 2004; accepted January 14, 2004.

 Disclosure: J. Li, None; N. Morlet, None; J.Q. Ng, None; J.B. Semmens, None; M.W. Knuiman, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Table 1.

 

 

Previously Published Incidence Rates of Endophthalmitis after Cataract Surgery

Table 1.

 

Previously Published Incidence Rates of Endophthalmitis after Cataract Surgery

Reference	Location	Period	Number of Procedures	Incidence (per 1000)	Notes
Population studies
Semmens et al.3 (2003)	Australia	1980–1998	94,653	1.99	*, †
Montan et al.6 (2002)	Sweden	1998	54,666	1.06	, ‡, †
Versteegh and Van7 (2000)	Netherlands	1996–1997	71,028	1.52	*
Wegener et al.8 (1998)	Denmark	1994–1995	951	0.00	, ‡
Norregaard et al.9 (1997)	Denmark	1985–1987	19,426	3.14	*, †
Javitt et al.10 (1994)	USA	1986–1987	57,103	0.77	*, §, ∥
Javitt et al.11 (1991)	USA	1984	324,032	1.36	*, §
Multicenter studies
Versteegh and Van7 (2000)	Netherlands	1996–1997	33,750	1.16	*, †
Schmitz et al.12 (1999)	Germany	1996	340,633	0.78	*, †
Desai et al.13 (1999)	United Kingdom	1997	15,787	1.65	*, †, ‡
Desai14 (1993)	United Kingdom	1990 (1 week)	998	3.00	, †, ‡
Fisch et al.15 (1991)	France	1988–1989	26,731	3.23	, †, ‡
Single center studies
Mayer et al.16 (2003)	United Kingdom	1991–2001	18,191	1.65	*, †
Riley et al.17 (2002)	New Zealand	2000	488	2.05	, †, ‡
Eifrig et al.4 (2002)	USA	1995–2001	21,972	0.36	*, †, ¶
Yorston et al.18 (2002)	Africa	1999	1,800	0.00	, ‡
Ernest et al.19 (2000)	Czech Republic	1990–1999	13,247	2.57
Colleaux and Hamilton20 (2000)	Canada	1994–1998	13,886	0.72	*, †
Bohigian21 (1999)	USA	1985–1996	19,269	0.62	*, †
Aaberg et al.2 (1998)	USA	1990–1994	18,530	0.92	*, ¶, #
Somoni et al.5 (1997)	Canada	1989–1996	13,285	2.41	*, †
Chitkara and Smerda22 (1997)	United Kingdom	1987–1991	1,552	1.93	*
Stanila23 (1996)	Romania	1990–1995	759	3.95
de Gottrau and Leuenberger24 (1994)	Switzerland	1983–1992	6,954	1.15	*
Kattan et al.25 (1991)	USA	1984–1989	23,625	0.72	*, ¶, #
Kattan et al.25 (1991)	USA	1976–1982	7,552	0.93	*, ¶, #
Fahmy26 (1975)	Denmark	1964–1974	4,498	5.33	*, †
Allen and Mangiaracine27 (1974)	USA	Not stated	36,000	0.86	, †
Christy and Lall28 (1973)	Pakistan	1957–1972	77,093	4.96	, †

 Studies were not included if cataract surgery numbers were not provided to enable calculation of cumulative incidence.

* Retrospective designs.

† Endophthalmitis defined clinically.

‡ Prospective design.

§ Endophthalmitis defined according to International Classification of Diseases codes and not validated clinically.

∥ Study population consisted of a 5% random population sample.

¶ Studies originated from the same center.

# Endophthalmitis defined as culture positive microbiological specimens.

Table 2.

 

 

Characteristics of Cataract Surgery Patients and Multivariate Logistic Regression Analysis of Risk Factors for Postcataract Surgery Endophthalmitis

Table 2.

 

Characteristics of Cataract Surgery Patients and Multivariate Logistic Regression Analysis of Risk Factors for Postcataract Surgery Endophthalmitis

Variable	Cases of Postoperative Endophthalmitis [No. (%)]	Others— Cataract Procedures [No. (%)]	Incidence of Endophthalmitis (Per 1,000 Cataract Operations)	OR	95% CI	P
Patient factors
Gender
Male (reference)	83 (39.5)	49,029 (42.0)	1.69	1.00
Female	127 (60.5)	67,844 (58.0)	1.87	1.12	0.84–1.49	0.44
Age
<80 years (reference)	133 (63.3)	83,978 (71.9)	1.58	1.00
≥80 years	77 (36.7)	32,895 (28.1)	2.34	1.50	1.13–1.99	0.005
Cancer history
No cancer or previous skin cancer (reference)	199 (94.8)	111,985 (95.8)	1.77	1.00
Previous cancer (excluding skin cancer)	11 (5.2)	4,888 (4.2)	2.25	1.23	0.67–2.26	0.50
Charlson comorbidity
None (reference)	189 (90.0)	104,254 (89.2)	1.81	1.00
Present	21 (10.0)	12,619, † (10.8)	1.66	1.04	0.66–1.65	0.86
Health insurance status
Private (reference)	123 (58.6)	64,825 (55.5)	1.89	1.00
Public	84 (40.0)	50,600 (43.3)	1.66	1.47	0.83–2.63	0.19
Other	3 (1.4)	1,448 (1.2)	2.07	0.77	0.24–2.53	0.67
Locality of residence*
Metropolitan (reference)	160 (76.2)	95,059 (81.3)	1.68	1.00
Rural	42 (20.0)	17,861 (15.3)	2.35	1.49	0.97–2.28	0.07
Remote	8 (3.8)	3,594 (3.1)	2.22	1.33	0.61–2.90	0.48
Hospital factors
Type
Public hospital (reference)	87 (41.4)	55,465 (47.5)	1.57	1.00
Private hospital	123 (58.6)	61,408 (52.5)	2.00	2.38	1.32–4.27	0.004
Location
Metropolitan hospital (reference)	182 (86.7)	103,383 (88.5)	1.76	1.00
Rural hospital	28 (13.3)	13,490 (11.5)	2.07	0.94	0.56–1.59	0.83
Total volume of cataract operations, †
The largest (24,301 procedures, n = 1) (reference)	39 (18.6)	24,262 (20.8)	1.60	1.00
Large (2,004–10,017 procedures, n = 16)	133 (63.3)	72,285 (61.8)	1.84	1.37	0.88–2.11	0.16
Medium (1,411–1,907 procedures, n = 7)	17 (8.1)	11,748 (10.1)	1.44	0.93	0.52–1.66	0.80
Small (<1,000 procedures, n = 28)	21 (10.0)	8,578 (7.3)	2.44	1.40	0.73–2.69	0.31
Length of stay
Same day, ‡	91 (43.3)	47,198 (40.4)	1.92	1.42	0.95–2.13	0.09
1 day	27 (12.9)	16,516 (14.1)	1.63	1.15	0.72–1.85	0.55
2–5 days (reference)	66 (31.4)	43,123 (36.9)	1.53	1.00
6–7 days	7 (3.3)	4,937 (4.2)	1.42	0.84	0.37–1.92	0.68
8 + days	19 (9.1)	5,099 (4.4)	3.71	2.08	1.18–3.67	0.01
Surgical factors
Technique
Extracapsular/Phacoemulsification (reference)	165 (78.6)	100,258 (85.8)	1.64	1.00
Intracapsular	15 (7.1)	5,145 (4.4)	2.91	1.67	0.95–2.95	0.08
Other cataract extraction/Lens-related procedures	30 (14.3)	11,470 (9.8)	2.61	1.52	0.95–2.45	0.08
Other eye procedures during the same admission, §
No other eye procedure (reference)	190 (90.5)	111,156 (95.0)	1.71	1.00
Glaucoma/drainage	4 (1.9)	1,913 (1.6)	2.09	1.11	0.42–2.95	0.83
Vitreoretinal procedure, ∥	6 (2.9)	1,190 (1.0)	5.02	2.71	1.18–6.23	0.02
Anterior vitrectomy	2 (0.9)	778 (0.7)	2.56	1.48	0.36–6.15	0.59
Corneal graft	1 (0.5)	393 (0.3)	2.54	1.14	0.16–8.33	0.90
Lacrimal/eyelid procedures	4 (1.9)	90 (0.1)	42.55	23.50	8.50–64.98	<0.001
Other	3 (1.4)	1,519 (1.3)	1.97	1.05	0.33–3.28	0.94
Environmental factors
Year of surgery (1980–2000)				0.98	0.95–1.02	0.27
Season of surgery
Spring (reference)	42 (20.0)	30,238 (25.9)	1.39	1.00
Summer	44 (21.0)	25,086 (21.4)	1.75	1.27	0.83–1.93	0.27
Autumn	61 (29.0)	30,941 (26.5)	1.97	1.41	0.95–2.08	0.08
Winter	63 (30.0)	30,608 (26.2)	2.05	1.48	1.00–2.18	0.05

 Cases = 210, Others = 116,873, Overall Incidence = 1.79 per 1,000. Standard errors were adjusted for clustering of procedures on individual patients.

* There were 359 (0.3% of 117,083) procedures missing a locality but none were cases.

† n denotes the number of hospital.

‡ Due to the virtual nonexistence of same-day surgery before 1990, additional analysis restricted to procedures performed in 1990–2000 revealed that compared with inpatient admission for 2–5 days, the odds ratio for endophthalmitis after same-day surgery was 1.62 (95% CI 1.01–2.61, P = 0.047).

§ Some of the other eye procedures overlap, with some patients having more than one other eye procedure during the same admission.

∥ Data include procedures to treat endophthalmitis during the same admission as the cataract procedure.

Figure 1.

 

(A) Average length of stay and percentage of same-day cataract procedures in Western Australia. (B) Changing patterns of length of stay after cataract surgery pre- and post-1990 in Western Australia.

Figure 2.

 

Trends in endophthalmitis following extracapsular and phacoemulsification procedures in Western Australia.

Table 3.

 

 

Logistic Regression Analysis of the Influence of Surgical Type on the Risk of Postoperative Endophthalmitis

Table 3.

 

Logistic Regression Analysis of the Influence of Surgical Type on the Risk of Postoperative Endophthalmitis

Variable	Univariate Association* (Model 1)					Association Adjusted for Demographic Variables and Hospital, † (Model 2)					Association Adjusted for all Variables in the Multivariate Model, ‡ (Model 3)
		OR	95% CI	P	OR	95% CI	P	OR	95% CI	P
Extracapsular/phacoemulsification (reference)	1.00	—	—	1.00	—	—	1.00	—	—
Intracapsular	1.81	1.06, 3.10	0.03	1.80	1.06, 3.09	0.03	1.67	0.95, 2.95	0.08
Other cataract and lens-related procedures	1.64	1.06, 2.54	0.03	1.62	1.04, 2.54	0.03	1.52	0.95, 2.45	0.08
Year of surgery	1.01	0.98, 1.03	0.70	1.00	0.97, 1.03	0.99	0.98	0.95, 1.02	0.27

 Cases = 210 and Others = 116,873. Standard errors were adjusted for clustering of procedures on individual patients.

* Model 1: Surgical technique only.

† Model 2: Model 1 plus age, gender, locality of patient residence, and hospital type (public or private).

‡ Model 3: All variables in the multivariate logistic regression model 2 .

Table 4.

 

 

Endophthalmitis Risk Scenarios Based on Actual Data and the Multivariate Logistic Regression Model

Table 4.

 

Endophthalmitis Risk Scenarios Based on Actual Data and the Multivariate Logistic Regression Model

Actual Data*	Scenario, †	Estimated Number of Cases	Estimated Incidence Rate, ‡	Difference in Incidence Rate, § (%)
Age < 80
Cataract procedures 84,111	Same-day surgery in a private hospital in winter	144	1.73	+8
Endophthalmitis cases 133	Inpatient 2–5 days in a private hospital in winter	101	1.20	−24
Incidence rate 1.58, ‡	Inpatient 2–5 days in a public hospital in winter	43	0.51	−68
	Inpatient 2–5 days in a public hospital in spring	29	0.34	−78
Age ≥ 80
Cataract procedures 32,972	Same-day surgery in a private hospital in winter	84	2.56	+10
Endophthalmitis cases 77	Inpatient 2–5 days in a private hospital in winter	59	1.80	−23
Incidence rate 2.34, ‡	Inpatient 2–5 days in a public hospital in winter	25	0.76	−67
	Inpatient 2–5 days in a public hospital in spring	17	0.51	−78
All patients
Cataract procedures 117,083	Inpatient 2–5 days in a public hospital during:
Endophthalmitis cases 210	Winter	68	0.58	−68
Incidence rate 1.79, ‡	Spring	46	0.39	−78

* Actual number of procedures performed and cases occurring in Western Australia in 1980–2000 for each respective group.

† Calculated assuming that all other risk factors in 2 are at the reference level.

‡ Number of cases per 1,000 procedures.

§ Percentage differences between actual and estimated incidence rates.