- Email: [email protected]
Endophthalmitis Caused by the Coagulase-negative Staphylococci
Endophthalmitis Caused by the Coagulase--negative Staphylococci 2. Factors Influencing Presentation after Cataract Surgery L. David Ormerod, MD, MSC,l,2,3 Lynne E. Becker, BS, MSPH, 4 Robert]. Cruise, PhD,4 H. Irene Grohar, MS,3 Barbara G. Paton, BS, MT,s Albert R. Frederick, Jr., MD,6,7 Trexler M. Topping, MD,6,7 John]. Weiter, MD, PhD,I,6,7 Sheldon M. Buzney, MD,l,6,7 Ann S. Baker, MD 9,lO Purpose: This study, comprising 60 patients with coagulase-negative staphylococcal endophthalmitis which occurred after cataract surgery, was designed to define the variation in disease presentation and visual outcome and to evaluate statistically the role of the primary surgery and its management. Methods: An intensive evaluation of microbiological, inpatient, outpatient, and cataract surgery charts was made retrospectively using a standardized protocol. The predictive value of surgical, iatrogenic, and clinical factors was analyzed for their influence on defined aspects of the disease pattern and of the visual results using multiple regression models, via a stepwise technique. Results: There was commonly a significant asymptomatic latent period after cataract surgery. The median diagnostic delay was 7 days; 22% of patients presented after 2 weeks and 12% after 1 month. Symptoms progressed longer than 3 days in 25% of patients. Ten percent had no pain. Clinical variation proved largely unrelated to cataract surgery events and postoperative management; bacterial factors were implicated. Good visual outcome was associated statistically with intensive topical corticosteroid in the symptomatic period, but was negatively associated with operative subconjunctival corticosteroid. Conclusions: The clinical variation in cases of postoperative coagulase-negative staphylococcal endophthalmitis poses particular problems for diagnosis in the outpatient setting. Surgical and perioperative events (except corticosteroid use) probably can be disregarded in studies of endophthalmitis management. Ophthalmology 1993;100:724-729
Originally received October 6, 1992. Revision accepted: December 8, 1992. 1 Schepens Eye Research Institute, Harvard University Medical School, Boston. 2
Department of Ophthalmology, Lorna Linda University, Lorna Linda.
3
Currently affiliated with Kresge Eye Institute, Detroit.
Department of Biostatistics and Epidemiology, School of Public Health, Lorna Linda University, Lorna Linda.
4
Porter Bacteriology Laboratory, Massachusetts Eye and Ear Infirmary, Harvard University Medical School, Boston. 6 Department of Ophthalmology, Massachusetts Eye and Ear Infirmary, Harvard University Medical School, Boston. 7 Ophthalmic Consultants of Boston and Centers for Eye Research, Boston. 5
724
Retina Associates, Boston. Infectious Diseases Unit, Department of Medicine, Massachusetts General Hospital, Harvard University Medical School, Boston. 10 Infectious Disease Service, Massachusetts Eye and Ear Infirmary, Harvard University Medical School, Boston. Presented in part at the American Academy of Ophthalmology Annual Meeting, Oct/Nov 1989, New Orleans, at the American Society of Cataract and Refractive Surgery Annual Meeting, March 1990, Los Angeles, and at the Association for Research in Vision and Ophthalmology Annual Meeting, April 1992, Sarasota. Correspondence to L. David Ormerod, MD, Vitreoretinal Unit, Department of Ophthalmology, Kresge Eye Institute, 4717 St. Antoine, Detroit, MI 4820 I. Reprint requests to Ann S. Baker, MD, Infectious Disease Service, Massachusetts Eye and Ear Infirmary, 243 Charles St, Boston, MA 02114.
8
9
Ormerod et al . Postoperative Coagulase-negative Staphylococcal Endophthalmitis Although the incidence rates of postoperative infection gradually decline, I the annual expansion in the numbers of cataract extractions performed increases the prevalence of postoperative endophthalmitis. Twenty years ago, infected eye surgeries invariably cultured major pathogens,2 and an acute fulminant disease developed during the postoperative hospital stay. Recent large-scale series have recorded the emergence of the coagulase-negative staphylococci (Staphylococcus epidermidis, Staphylococcus haemolyticus, Staphylococcus warneri, Staphylococcus hominis, Staphylococcus capitis, and five other species) as the predominant causes of postoperative endophthalmitis, with isolation rates of approximately 3S% to 4S%.3-5 The disease pattern in this infection is often atypical and delayed, engendering diagnostic problems from the perspective of outpatient postoperative care. Sixty cases of coagulase-negative staphylococcal endophthalmitis occurring after cataract surgery were analyzed to give a detailed clinical description of the disease process. To identify causes for the observed variation in disease patterns, potential predictive factors were assessed, using multiple regressions, for their influence on defined disease characteristics and visual outcome.
Subjects and Methods Sixty cases of coagulase-negative staphylococcal endophthai mitis occurring after cataract surgery were identified from a retrospective, unselected series of90 patients with coagulase-negative endophthalmitis described in a companion article. 6 The assessment and management of all patients were undertaken at the Massachusetts Eye and Ear Infirmary in Boston. The cases were defined by the isolation of the requisite organism(s) from intraocular specimens in the context of a clinical diagnosis of postcataract endophthalmitis, using a restrictive case definition. The survey and microbiologic methods have been outlined above. 6 Almost all patients received a pars plana vitrectomy, intraocular antibiotics, and a 7- to lO-day course of intravenous antibiotics. Visual outcome was de-
fined as the best-corrected visual acuity at least 6 months after definitive treatment. 6 An attempt was made to identify statistically significant predictor variables for final visual acuity and also to investigate the contribution of surgical, iatrogenic, clinical, and bacterial factors to the remarkable variation of disease pattern observed. Multiple regression models7 were developed using SPSS/PC+ software (SPSS, Chicago, IL, 1990). The four independent variables of interest are shown in Table I. For each independent variable, potential dependent variables (Table 2) were entered into exploratory multiple regressions in a stepwise procedure,? with entry criteria of P s; 0.2 and exit criteria of P s; 0.3. In the final models, criteria were restricted to variables with a significance of P s; O.OS. The strengths of the ensuing formulas were assessed 7 by estimation of the multiple correlation coefficient (R), by the squared multiple correlation coefficient (R 2), and by the adjusted R2.
Results Preceding Events at Cataract Surgery Twenty-four cases of intracapsular cataract extraction (ICCE), three cases of secondary anterior chamber (AC) intraocular lens (I0L) insertions, and 33 cases of extracapsular cataract extraction (ECCE) comprise the 60 patients with coagulase-negative endophthalmitis after cataract surgery. Triple surgeries, as well as vitreous wick and inadvertent bleb infections, were excluded. Intraocular lenses were involved in 40 of the patients, S with ICCE, 32 with ECCE, and 3 with secondary AC IOL insertions after ICCE. Investigation and management were relatively uniform over the study period. 6 Seventy-nine percent (38 of 48 patients with information) had received prophylactic topical antibiotics before cataract surgery, using many regimens. Nineteen percent of the cataract patients with preoperative prophylactic antibiotics were infected with resistant organisms. During the operative field preparation, more than
Table 1. Independent Variables in the Statistical Modeling of Disease Character and Outcome in Coagulase-negative Staphylococcal Endophthalmitis Independent Variables
Length of presymptomatic period Duration of endophthalmitis symptoms to diagnosis Painless endophthalmitis' Final (>6 mos) visual acuityt CF
=
counting fingers; NLP
=
Stratification
4 levels: days 1-3; days 4-7; wks 2-4; >1 mo 5 levels: days 0-2; days 3-4; days 5-7; wks 2-4; >1 mo 5 levels: 20/20-20/30; 20/40-20/50; 20/60-20/80; 20/100-20/400; CF-NLP
no light perception.
• Disregarding four patients with minimal discomfort.
t Patients with intercedent causes of visual loss excluded from consideration for the purposes of statitistical analyses.
725
Ophthalmology
Volume 100, Number 5, May 1993
Table 2. Dependent Variables Used in the Statistical Modeling of the Four Independent Variables in Table 1 Events at Cataract Surgery Type of cataract surgery: ECCE, ICCE, 2° IOL IOL or no IOL Topical antibiotic prophylaxis Subconjunctival corticosteroid Subconjunctival antibiotic Major operative complications: vitreous loss, vitrectomy, broken capsule Presymptomatic and Symptomatic Phases Duration of presymptomatic period (stratified 4 levels)* Duration of endophthalmitis symptoms to diagnosis (stratified 5Ievels)* Presence/absence of pain* Prediagnostic use of intensive topical corticosteroids (>4 times daily) Prediagnostic use of oral corticosteroid Endophthalmitis Assessment and Managementt Hypopyon Retinal reflex at diagnosis (retinal view, red reflex, no red reflex) Retinal viewing at vitrectomy (full view, partial view, no view) Cataract surgery wound repair at vitrectomy Oral corticosteroid Intravitreal corticosteroid Staphylococcus epidermidis or nonepidermidis isolate Methicillin resistance of isolate ECCE = extracapsular cataract extraction; ICCE = intracapsular cataract extraction; 2 0 IOL = secondary intraocular lens insertion .
conjunctival antibiotic injection, mainly gentamicin (2040 mg). Only 33% of the patients given subconjunctival antibiotics at cataract surgery were infected with resistant organisms. Forty-six percent (23 of 50) of patients had received a perioperative subconjunctival corticosteroid injection. Three patients were receiving oral corticosteroids. Clinical Pattern of Coagulase-negative Staphylococcal Endophthalmitis after Cataract Surgery Diagnosis frequently was delayed from the time of cataract surgery (Fig 1). Seven percent (4 of 59 patients with information) were diagnosed within 48 hours of cataract surgery and 58% (34 patients) were diagnosed within 1 week. Twenty percent (12 patients) were diagnosed in the second week, 10% (6 patients) in the third and fourth postoperative weeks, 7% (4 patients) in the second month, and the remaining three patients (5%) more than 8 weeks after the surgery. The median diagnostic delay was 7 days. Twenty-eight percent (17 of 60) of the patients underwent revision of the cataract wound at the time of endophthalmitis surgery; a positive Seidel test was detected in only 5% (3 patients). Symptoms of pain and/or visual failure often developed gradually and progressively over a few days (Fig 2). The duration of symptoms was more than 48 hours in 40% (23 of 57) of the patients, more than 72 hours in 26% (15 patients), and more than 1 week in 11 % (6 patients). The disease was painless in 10% (6 patients). No differences were discerned in the clinical patterns of ECCE and ICCE infections.
• Factors can be dependent variables in analyses in which they are not the subject of the study (Le., the independent variable).
t The relative homogeneity of treatment protocols regarding vitrectomy, intravitreal antibiotics and intravenous antibiotics precludes their discriminant analysis. Note: Most but not all the dependent variables were incorporated in each of the statistical models. For example, endophthalmitis treatment variables would not be relevant to investigation of the presymptomatic and prodromal phases.
18 16
(/)
c
14
a>
•.;:J
Cll
a..
half the patients had either no operative conjunctival antisepsis or only saline irrigation; almost equal numbers of the remaining patients were given either a povidone-iodine preparation or a mild silver protein solution. The favored operative lid preparations were povidone-iodine, liquid soap-alcohol, and chlorhexidine ± alcohol (7:2: 1, proportionally). Twenty-seven percent (16 of 60) of the patients in whom coagulase-negative staphylococcal endophthalmitis developed had experienced a major operative complication, principally vitreous loss; broken posterior capsule without vitreal disturbance (ECCE) and primary wound leak also occurred. Operative complications had occurred in 37% of the ICCEs and 19% of the ECCEs. Forty-eight percent of the cataract patients (24 of 50 with information) had been given a perioperative sub-
726
'0
a>
0)
Cll
c
a> u .... a>
a..
o
1
2
3
4
5
Day
6
7
8- 11- 3 10 14
4
Week
2
3
>3
Month
Time: Cataract Surgery to Diagnosis Figure 1. Interval from time of surgery to diagnosis in 59 patients with coagulase-negative staphylococcal endophthalmitis after cataract surgery; the date of cataract surgery was not known for one patient.
Ormerod et al . Postoperative Coagulase-negative Staphylococcal Endophthalmitis portance of hypopyon in the natural history was supported statistically. Length of Symptomatic Period before Diagnosis. There were no significant associations with the duration of the symptomatic prodrome to diagnosis. Chronic endophthalmitis patterns were unrelated to surgical and iatrogenic factors. Presence or Absence of Pain. No statistically significant associations were found; notably, there was no causal relationship with corticosteroid use. Visual Outcome. Investigating predictors of final visual acuity, we ascertained a mathematical model that statistically can predict the visual outcome approximately 38% of the time:
30
(/)
c
Q)
~
(L
o
Q)
~
c
Q)
<.> .... Q)
(L
Day
Week
Month
Visual acuity
Visual Outcome Long-term visual results are known for 51 cataract surgery patients; the visual outcomes after ECCE and ICCE are shown separately in Figure 3. Overall, 49% of these 51 patients achieved at least 20/50 visual acuity, 55% (28 patients) at least 20170, and 69% (35 patients) at least 20/ 400; 6% (3 patients) lost light perception. There was a noticeably poorer prognosis after vitreous loss in ICCE. When patients with intercurrent causes of visual loss were excluded, 57% (25 of 44) of post-cataract patients achieved a visual acuity of 20/50 or better.
63.77
+ 11.86
X (prediagnostic intensive topical corticosteroid)
Symptom Duration to Diagnosis Figure 2. Duration of clinical symptoms before diagnosis of coagulasenegative endophthalmitis in 57 patients after cataract surgery; the time of onset of symptoms was not recorded in three patients.
=
- 36.84 X (operative subconjunctival corticosteroid) There is a relatively strong correlation, R = 0.63, with R2 of 0.4 and an adjusted R2 of 0.38. The mUltiple regression and correlation produced an F-ratio equal to 18.76 with a significance of P < 0.001. Numerous factors did not prove to be significant predictors of visual outcome. These included the use of prophylactic antibiotics at cataract surgery, ECCE or ICCE, the presence, absence, or type of IOL, status of the posterior capsule, occurrence of major cataract surgery complications, the necessity for wound repair at vitrectomy, duration of the presymptomatic period, length of the inflammatory prodrome before diagnosis, pain free disease history, retinal reflex at diagnosis, or the methicillin sensitivity of the bacterial isolate.
25
Statistical Modeling of Factors Defining Disease Pattern and Visual Outcome The independent and dependent variables used in this statistical analysis are shown in Tables 1 and 2. Duration of Presymptomatic Interval. A weak model (10% predictability) was identified, influencing delays in disease appearance: Presymptomatic interval
=
1.98
+ 0.63 X (hypopyon)
- 0.35 X (preoperative topical antibiotic prophylaxis) There is a weak correlation coefficient, R of 0.36, with R2 of 0.13, and an adjusted R2 of 0.10. The multiple regression and correlation produced an F-ratio of 4.13 with a significance of P < 0.02. In particular, there was no statistically significant association with the presence or absence of an IOL, the type of cataract surgery, cataract surgery complications, the posterior capsule status, the use of corticosteroids, or with visual outcome. The im-
•
(/)
C Q)
o
20
ICCE n.23 ECCE n.28
~
(L
'0
15
Q)
Cl ell
C Q) <.> .... Q) a..
10 5 0 Final Visual Acuity
Figure 3. Best-corrected visual acuities of 51 patients with coagulasenegative endophthalmitis occurring after cataract surgery; visual acuities are recorded at least 6 months after endophthalmitis treatment. The data are presented separately for extracapsular cataract extractions (ECCE) (n = 28) and intracapsular cataract extractions (ICCE) (n = 23). A final visual acuity was not available for 9 of the 60 patients.
727
Ophthalmology
Volume 100, Number 5, May 1993
Discussion The findings of this series contrast with the disease pattern of endophthalmitis caused by more virulent bacteria that presents characteristically over the first 72 hours. 8 The considerable variation in the delay from cataract surgery to diagnosis (range, 1 day to 28 weeks; median, 7 days; mean, 21 days) emphasizes the atypical pattern of endophthalmitis caused by the coagulase-negative staphylococci. There was a mean diagnostic interval of 7 days for the 22 patients with postsurgical coagulase-negative staphylococcal endophthalmitis described by Bode et al 9 and of 15 days for 13 such patients reported by Ficker et al.lO Forty-two percent of our patients presented more than 1 week, 22% more than 2 weeks, and 12% more than 1 month after surgery. We recorded a 29% prevalence of cataract wound abnormalities coincident with the diagnosis of endophthalmitis compared with a 22% prevalence reported by Driebe et al4 in a nonselected endophthalmitis series, although an actual wound leak was discerned in only 5% of our series, with 11 % similarly affected in Driebe et aI's patients. We do not believe that a poorly healed wound without aqueous humor drainage, in the presence of postoperative inflammation, implies that a wound leak has been the source of infection, although wound revision should be undertaken in such cases. Most of the delay in diagnosis was accounted for by a prolonged asymptomatic interval. 6,9 The variable progression of endophthalmitis in the interval between symptoms and diagnosis is also a significant observation. Over one quarter of the cataract surgery patients gave a history of symptoms lasting more than 72 hours before diagnosis and one tenth had symptoms for more than 7 days, and occasionally for much longer. Painless endophthalmitis II occurred in 10% of the cataract patients; this entity deserves much greater recognition. Overall, 49% of the cataract patients achieved visual acuity of ~20/50, compared with 85%, 55%, 38%, and 33% in the post-cataract coagulase-negative staphylococcal endophthalmitis cases ofO'Day et aI, 12 Bode et al,9 Ficker et aI, 13 and Bohigian and Olk, 14 respectively. Severity and disease variables differ among series, Nearly one third of our cataract patients with endophthalmitis retained less than 20/400 visual acuity. The often slowly progressive and delayed nature of the physical signs and symptoms emphasizes the need for immediate suspicion in patients with postoperative inflammation that is out of the ordinary. Improved familiarity with the spectrum of endo phthalmitis caused by these organisms (and other lesser pathogens) should enhance the management of delayedonset, subacute, chronic, and painless postoperative endophthalmitis. 5,9,10,12,13,15,16 Multiple regression models investigated the role ofnumerous iatrogenic and clinical factors (Table 2) in modeling the disease pattern, cognizant that bacterial virulence and host reactivity remained unmeasured. The clinical variation proved largely unrelated to details of the cataract surgery and postoperative management. The negative association between the asymptomatic period and topical
728
antibiotic prophylaxis suggests, perhaps, some role for topical antibiotic selection for more virulent bacteria. No predictive model could explain the often extended clinical course in these patients. In particular, neither the interval before symptoms developed nor the course of the symptoms was influenced significantly by major cataract surgery complications, the integrity of the lens capsule, the use of perioperative subconjunctival corticosteroids and/ or antibiotics, or the necessity for subsequent wound repair. These data support the premise that the use of perioperative corticosteroidsl 3 and antibiotics9 mask this disease, rather than delay its onset or alter fundamentally the disease process. We were unable to find evidence that the status of the posterior capsule was important in the disease spectrum. No significant differences existed in the clinical patterns between cases of ECCE or ICCE, with or without an IOL. It remains to be determined whether the enhanced adhesion of many coagulase-negative staphylococcal strains to the biomaterial I 7-20 (via extracellular bacterial biofilm 21 ,22) might be a pathogenetic factor in postoperative endophthalmitis. Painless endophthalmitis had no significant association with corticosteroid use. II ,23 The disease spectrum and outcome in coagulase-negative staphylococcal endophthalmitis are probably mainly due, therefore, to intrinsic differences in bacterial pathogenicity. The size and site of bacterial inoculation and varying bacterial virulence are likely to be important factors. 24,25 An intriguing and unexpected finding was the relatively strong direct association of a good visual outcome (protective effect) with the use of intensive topical corticosteroids in the symptomatic period before diagnosis. Paradoxically, the use of subconjunctival corticosteroids at cataract surgery, coincidental with the bacterial inoculation, was negatively correlated with outcome; this novel observation suggests that the suppression of early inflammatory events might modify adversely subsequent hostbacterial inflammatory and immunologic relationships in coagulase-negative staphylococcal endophthalmitis. Of particular interest was the failure to detect a significant association between visual outcome and such variables as the type of cataract surgery, major operative complications, the integrity of the lens capsule, subconjunctival antibiotic prophylaxis, endophthalmitis treatment with oral or intravitreal corticosteroids (confounded by associated vitrectomy and antibiotic administration), the asymptomatic and symptomatic periods, the retinal reflex at diagnosis, and the methicillin sensitivity of the coagulase-negative staphylococcal isolate. No inferences as to the effectiveness of treatment modalities could be made from these data because of management uniformity. This retrospective study suffers from the flaws inherent in any such investigation. However, our exploratory work approaches questions that have not been addressed previously, and multiple regression models provide a rigorous analysis of comprehensive data. The relative lack of influence of the multiplicity of peri operative variables in cataract surgery on the subsequent coagulase-negative en-
Ormerod et al . Postoperative Coagulase-negative Staphylococcal Endophthalmitis dophthalmitis suggests that, with the exception of corticosteroid use, they probably can be disregarded in the analysis of prospective studies of endophthalmitis management. Acknowledgments The authors thank the staff ophthal-
mologists at the Massachusetts Eye and Ear Infirmary, Boston, and the 39 cataract surgeons from across New England for their active contributions.
References 1. Kattan HM, flynn HW Jr, Pflugfelder SC, et al. Nosocomial endophthalmitis survey: current incidence of infection after intraocular surgery. Ophthalmology 1991 ;98:227-38. 2. Postoperative infections of the eye. In: Locatcher-Khorazo 0, Seegal BC, eds. Microbiology of the Eye. St. Louis: CV Mosby, 1972:77-85. 3. Rowsey JJ, Newsom DL, Sexton OJ, Harms WK. Endophthalmitis: current approaches. Ophthalmology 1982;89: 1055-66. 4. Driebe WT Jr, Mandelbaum S, Forster RK, et al. Pseudophakic endophthalmitis: diagnosis and management. Ophthalmology 1986;93:442-8. 5. Shrader SK, Band JD, Lauter CB, Murphy P. The clinical spectrum of endophthalmitis: incidence, predisposing factors, and features influencing outcome. J Infect Dis 1990;162:115-20. 6. Ormerod LD, Ho DO, Becker LE, et al. Endophthalmitis caused by the coagulase-negative staphylococci. 1. Disease spectrum and outcome. Ophthalmology 1993; 100:715-723. 7. Keppel G, Zedeck S. Data Analysis for Research Design: Analysis-of-Variance and Multiple Regression/Correlation Approaches. New York: WH Freeman, 1989. 8. Jaffe NS, Jaffe MS, Jaffe GF. Cataract Surgery and Its Complications, 5th ed. St Louis: CV Mosby, 1990:506-42. 9. Bode DO Jr, Gelender H, Forster RK. A retrospective review of endophthalmitis due to coagulase-negative staphylococci. Br J Ophthalmol 1985;69:915-9. 10. Ficker LA, Meredith TA, Wilson LA, Kaplan HJ. Role of vitrectomy in Staphylococcus epidermidis endophthalmitis. Br J Ophthalmol 1988;72:386-9. 11. Deutsch T A, Goldberg MF. Painless endophthalmitis after cataract surgery. Ophthalmic Surg 1984;15:837-40. 12. O'Day OM, Jones DB, Patrinely J, Elliott JH. Staphylococcus epidermidis endophthalmitis: visual outcome fol-
13. 14. 15. 16.
17.
18. 19.
20. 21. 22.
23. 24. 25.
lowing noninvasive therapy. Ophthalmology 1982;89:35460. Ficker L, Meredith TA, Wilson LA, et al. Chronic bacterial endophthalmitis. Am J Ophthalmol 1987;103:745-8. Bohigian GM, Olk RJ. Factors associated with a poor visual result in endophthalmitis. Am J Ophthalmol 1986;101:33241. Fox GM, Joondeph BC, flynn HW Jr, et al. Delayedonset pseudophakic endophthalmitis. Am J Ophthalmol 1991;111: 163-73. Weber OJ, Hoffman KL, Thoft RA, Baker AS. Endophthalmitis following intraocular lens implantation: report of 30 cases and review of the literature. Rev Infect Dis 1986;8: 12-20. Christensen GO, Simpson W A, Bisno AL, Beachey EH. Adherence of slime-producing strains of Staphylococcus epidermidis to smooth surfaces. Infect Immun 1982;37:31826. Hogt AH, Dankert J, de Vries JA, Feijen J. Adhesion of coagulase-negative staphylococci to biomaterials. J Gen Microbiol 1983; 129:2959-68. Herrmann M, Vaudaux PE, Pittet 0, et al. Fibronectin, fibrinogen, and laminin act as mediators of adherence of clinical staphylococcal isolates to foreign materials. J Infect Dis 1988;158:693-701. Griffiths PG, Elliott TSJ, McTaggart L. Adherence of Staphylococcus epidermidis to intraocular lenses. Br J Ophthalmol 1989;73:402-6. Costerton JW, Cheng K-J, Geesey GG, et al. Bacterial biofilms in nature and disease. Annu Rev Microbiol 1987;41: 435-64. Hartmann C, Jansen B, Schumacher-Perdreau E, Peters G. Bakterielle Spatendophthalmitis nach Kunstlinsenimplantation. Eine chronische, IOL-assoziierte "Plastikinfektion." In: Schott K, Jacobi KW, Freyler H, eds. 4. Kongrel3 der Deutschen Gesellschaft fUr Intraokularlinsen Implantation: 6. bis 7. April 1990, Essen, Berlin: Springer-Verlag, 1991: 170-8. Laatikainen L, Tarkkanen A. Early vitrectomy in the treatment of post-operative purulent endophthalmitis. Acta Ophthalmol 1987;65:455-60. Ormerod LD, Edelstein MAC, Schmidt GJ, et al. The intraocular environment and experimental anaerobic bacterial endophthalmitis. Arch Ophthalmol 1987;105:1571-5. Shockley RK, Jay WM, Fishman PH, et al. Effect ofinoculum size on the induction of endophthalmitis in aphakic rabbit eyes. Acta Ophthalmol 1985;63:35-8.
729
Originally received October 6, 1992. Revision accepted: December 8, 1992. 1 Schepens Eye Research Institute, Harvard University Medical School, Boston. 2
Department of Ophthalmology, Lorna Linda University, Lorna Linda.
3
Currently affiliated with Kresge Eye Institute, Detroit.
Department of Biostatistics and Epidemiology, School of Public Health, Lorna Linda University, Lorna Linda.
4
Porter Bacteriology Laboratory, Massachusetts Eye and Ear Infirmary, Harvard University Medical School, Boston. 6 Department of Ophthalmology, Massachusetts Eye and Ear Infirmary, Harvard University Medical School, Boston. 7 Ophthalmic Consultants of Boston and Centers for Eye Research, Boston. 5
724
Retina Associates, Boston. Infectious Diseases Unit, Department of Medicine, Massachusetts General Hospital, Harvard University Medical School, Boston. 10 Infectious Disease Service, Massachusetts Eye and Ear Infirmary, Harvard University Medical School, Boston. Presented in part at the American Academy of Ophthalmology Annual Meeting, Oct/Nov 1989, New Orleans, at the American Society of Cataract and Refractive Surgery Annual Meeting, March 1990, Los Angeles, and at the Association for Research in Vision and Ophthalmology Annual Meeting, April 1992, Sarasota. Correspondence to L. David Ormerod, MD, Vitreoretinal Unit, Department of Ophthalmology, Kresge Eye Institute, 4717 St. Antoine, Detroit, MI 4820 I. Reprint requests to Ann S. Baker, MD, Infectious Disease Service, Massachusetts Eye and Ear Infirmary, 243 Charles St, Boston, MA 02114.
8
9
Ormerod et al . Postoperative Coagulase-negative Staphylococcal Endophthalmitis Although the incidence rates of postoperative infection gradually decline, I the annual expansion in the numbers of cataract extractions performed increases the prevalence of postoperative endophthalmitis. Twenty years ago, infected eye surgeries invariably cultured major pathogens,2 and an acute fulminant disease developed during the postoperative hospital stay. Recent large-scale series have recorded the emergence of the coagulase-negative staphylococci (Staphylococcus epidermidis, Staphylococcus haemolyticus, Staphylococcus warneri, Staphylococcus hominis, Staphylococcus capitis, and five other species) as the predominant causes of postoperative endophthalmitis, with isolation rates of approximately 3S% to 4S%.3-5 The disease pattern in this infection is often atypical and delayed, engendering diagnostic problems from the perspective of outpatient postoperative care. Sixty cases of coagulase-negative staphylococcal endophthalmitis occurring after cataract surgery were analyzed to give a detailed clinical description of the disease process. To identify causes for the observed variation in disease patterns, potential predictive factors were assessed, using multiple regressions, for their influence on defined disease characteristics and visual outcome.
Subjects and Methods Sixty cases of coagulase-negative staphylococcal endophthai mitis occurring after cataract surgery were identified from a retrospective, unselected series of90 patients with coagulase-negative endophthalmitis described in a companion article. 6 The assessment and management of all patients were undertaken at the Massachusetts Eye and Ear Infirmary in Boston. The cases were defined by the isolation of the requisite organism(s) from intraocular specimens in the context of a clinical diagnosis of postcataract endophthalmitis, using a restrictive case definition. The survey and microbiologic methods have been outlined above. 6 Almost all patients received a pars plana vitrectomy, intraocular antibiotics, and a 7- to lO-day course of intravenous antibiotics. Visual outcome was de-
fined as the best-corrected visual acuity at least 6 months after definitive treatment. 6 An attempt was made to identify statistically significant predictor variables for final visual acuity and also to investigate the contribution of surgical, iatrogenic, clinical, and bacterial factors to the remarkable variation of disease pattern observed. Multiple regression models7 were developed using SPSS/PC+ software (SPSS, Chicago, IL, 1990). The four independent variables of interest are shown in Table I. For each independent variable, potential dependent variables (Table 2) were entered into exploratory multiple regressions in a stepwise procedure,? with entry criteria of P s; 0.2 and exit criteria of P s; 0.3. In the final models, criteria were restricted to variables with a significance of P s; O.OS. The strengths of the ensuing formulas were assessed 7 by estimation of the multiple correlation coefficient (R), by the squared multiple correlation coefficient (R 2), and by the adjusted R2.
Results Preceding Events at Cataract Surgery Twenty-four cases of intracapsular cataract extraction (ICCE), three cases of secondary anterior chamber (AC) intraocular lens (I0L) insertions, and 33 cases of extracapsular cataract extraction (ECCE) comprise the 60 patients with coagulase-negative endophthalmitis after cataract surgery. Triple surgeries, as well as vitreous wick and inadvertent bleb infections, were excluded. Intraocular lenses were involved in 40 of the patients, S with ICCE, 32 with ECCE, and 3 with secondary AC IOL insertions after ICCE. Investigation and management were relatively uniform over the study period. 6 Seventy-nine percent (38 of 48 patients with information) had received prophylactic topical antibiotics before cataract surgery, using many regimens. Nineteen percent of the cataract patients with preoperative prophylactic antibiotics were infected with resistant organisms. During the operative field preparation, more than
Table 1. Independent Variables in the Statistical Modeling of Disease Character and Outcome in Coagulase-negative Staphylococcal Endophthalmitis Independent Variables
Length of presymptomatic period Duration of endophthalmitis symptoms to diagnosis Painless endophthalmitis' Final (>6 mos) visual acuityt CF
=
counting fingers; NLP
=
Stratification
4 levels: days 1-3; days 4-7; wks 2-4; >1 mo 5 levels: days 0-2; days 3-4; days 5-7; wks 2-4; >1 mo 5 levels: 20/20-20/30; 20/40-20/50; 20/60-20/80; 20/100-20/400; CF-NLP
no light perception.
• Disregarding four patients with minimal discomfort.
t Patients with intercedent causes of visual loss excluded from consideration for the purposes of statitistical analyses.
725
Ophthalmology
Volume 100, Number 5, May 1993
Table 2. Dependent Variables Used in the Statistical Modeling of the Four Independent Variables in Table 1 Events at Cataract Surgery Type of cataract surgery: ECCE, ICCE, 2° IOL IOL or no IOL Topical antibiotic prophylaxis Subconjunctival corticosteroid Subconjunctival antibiotic Major operative complications: vitreous loss, vitrectomy, broken capsule Presymptomatic and Symptomatic Phases Duration of presymptomatic period (stratified 4 levels)* Duration of endophthalmitis symptoms to diagnosis (stratified 5Ievels)* Presence/absence of pain* Prediagnostic use of intensive topical corticosteroids (>4 times daily) Prediagnostic use of oral corticosteroid Endophthalmitis Assessment and Managementt Hypopyon Retinal reflex at diagnosis (retinal view, red reflex, no red reflex) Retinal viewing at vitrectomy (full view, partial view, no view) Cataract surgery wound repair at vitrectomy Oral corticosteroid Intravitreal corticosteroid Staphylococcus epidermidis or nonepidermidis isolate Methicillin resistance of isolate ECCE = extracapsular cataract extraction; ICCE = intracapsular cataract extraction; 2 0 IOL = secondary intraocular lens insertion .
conjunctival antibiotic injection, mainly gentamicin (2040 mg). Only 33% of the patients given subconjunctival antibiotics at cataract surgery were infected with resistant organisms. Forty-six percent (23 of 50) of patients had received a perioperative subconjunctival corticosteroid injection. Three patients were receiving oral corticosteroids. Clinical Pattern of Coagulase-negative Staphylococcal Endophthalmitis after Cataract Surgery Diagnosis frequently was delayed from the time of cataract surgery (Fig 1). Seven percent (4 of 59 patients with information) were diagnosed within 48 hours of cataract surgery and 58% (34 patients) were diagnosed within 1 week. Twenty percent (12 patients) were diagnosed in the second week, 10% (6 patients) in the third and fourth postoperative weeks, 7% (4 patients) in the second month, and the remaining three patients (5%) more than 8 weeks after the surgery. The median diagnostic delay was 7 days. Twenty-eight percent (17 of 60) of the patients underwent revision of the cataract wound at the time of endophthalmitis surgery; a positive Seidel test was detected in only 5% (3 patients). Symptoms of pain and/or visual failure often developed gradually and progressively over a few days (Fig 2). The duration of symptoms was more than 48 hours in 40% (23 of 57) of the patients, more than 72 hours in 26% (15 patients), and more than 1 week in 11 % (6 patients). The disease was painless in 10% (6 patients). No differences were discerned in the clinical patterns of ECCE and ICCE infections.
• Factors can be dependent variables in analyses in which they are not the subject of the study (Le., the independent variable).
t The relative homogeneity of treatment protocols regarding vitrectomy, intravitreal antibiotics and intravenous antibiotics precludes their discriminant analysis. Note: Most but not all the dependent variables were incorporated in each of the statistical models. For example, endophthalmitis treatment variables would not be relevant to investigation of the presymptomatic and prodromal phases.
18 16
(/)
c
14
a>
•.;:J
Cll
a..
half the patients had either no operative conjunctival antisepsis or only saline irrigation; almost equal numbers of the remaining patients were given either a povidone-iodine preparation or a mild silver protein solution. The favored operative lid preparations were povidone-iodine, liquid soap-alcohol, and chlorhexidine ± alcohol (7:2: 1, proportionally). Twenty-seven percent (16 of 60) of the patients in whom coagulase-negative staphylococcal endophthalmitis developed had experienced a major operative complication, principally vitreous loss; broken posterior capsule without vitreal disturbance (ECCE) and primary wound leak also occurred. Operative complications had occurred in 37% of the ICCEs and 19% of the ECCEs. Forty-eight percent of the cataract patients (24 of 50 with information) had been given a perioperative sub-
726
'0
a>
0)
Cll
c
a> u .... a>
a..
o
1
2
3
4
5
Day
6
7
8- 11- 3 10 14
4
Week
2
3
>3
Month
Time: Cataract Surgery to Diagnosis Figure 1. Interval from time of surgery to diagnosis in 59 patients with coagulase-negative staphylococcal endophthalmitis after cataract surgery; the date of cataract surgery was not known for one patient.
Ormerod et al . Postoperative Coagulase-negative Staphylococcal Endophthalmitis portance of hypopyon in the natural history was supported statistically. Length of Symptomatic Period before Diagnosis. There were no significant associations with the duration of the symptomatic prodrome to diagnosis. Chronic endophthalmitis patterns were unrelated to surgical and iatrogenic factors. Presence or Absence of Pain. No statistically significant associations were found; notably, there was no causal relationship with corticosteroid use. Visual Outcome. Investigating predictors of final visual acuity, we ascertained a mathematical model that statistically can predict the visual outcome approximately 38% of the time:
30
(/)
c
Q)
~
(L
o
Q)
~
c
Q)
<.> .... Q)
(L
Day
Week
Month
Visual acuity
Visual Outcome Long-term visual results are known for 51 cataract surgery patients; the visual outcomes after ECCE and ICCE are shown separately in Figure 3. Overall, 49% of these 51 patients achieved at least 20/50 visual acuity, 55% (28 patients) at least 20170, and 69% (35 patients) at least 20/ 400; 6% (3 patients) lost light perception. There was a noticeably poorer prognosis after vitreous loss in ICCE. When patients with intercurrent causes of visual loss were excluded, 57% (25 of 44) of post-cataract patients achieved a visual acuity of 20/50 or better.
63.77
+ 11.86
X (prediagnostic intensive topical corticosteroid)
Symptom Duration to Diagnosis Figure 2. Duration of clinical symptoms before diagnosis of coagulasenegative endophthalmitis in 57 patients after cataract surgery; the time of onset of symptoms was not recorded in three patients.
=
- 36.84 X (operative subconjunctival corticosteroid) There is a relatively strong correlation, R = 0.63, with R2 of 0.4 and an adjusted R2 of 0.38. The mUltiple regression and correlation produced an F-ratio equal to 18.76 with a significance of P < 0.001. Numerous factors did not prove to be significant predictors of visual outcome. These included the use of prophylactic antibiotics at cataract surgery, ECCE or ICCE, the presence, absence, or type of IOL, status of the posterior capsule, occurrence of major cataract surgery complications, the necessity for wound repair at vitrectomy, duration of the presymptomatic period, length of the inflammatory prodrome before diagnosis, pain free disease history, retinal reflex at diagnosis, or the methicillin sensitivity of the bacterial isolate.
25
Statistical Modeling of Factors Defining Disease Pattern and Visual Outcome The independent and dependent variables used in this statistical analysis are shown in Tables 1 and 2. Duration of Presymptomatic Interval. A weak model (10% predictability) was identified, influencing delays in disease appearance: Presymptomatic interval
=
1.98
+ 0.63 X (hypopyon)
- 0.35 X (preoperative topical antibiotic prophylaxis) There is a weak correlation coefficient, R of 0.36, with R2 of 0.13, and an adjusted R2 of 0.10. The multiple regression and correlation produced an F-ratio of 4.13 with a significance of P < 0.02. In particular, there was no statistically significant association with the presence or absence of an IOL, the type of cataract surgery, cataract surgery complications, the posterior capsule status, the use of corticosteroids, or with visual outcome. The im-
•
(/)
C Q)
o
20
ICCE n.23 ECCE n.28
~
(L
'0
15
Q)
Cl ell
C Q) <.> .... Q) a..
10 5 0 Final Visual Acuity
Figure 3. Best-corrected visual acuities of 51 patients with coagulasenegative endophthalmitis occurring after cataract surgery; visual acuities are recorded at least 6 months after endophthalmitis treatment. The data are presented separately for extracapsular cataract extractions (ECCE) (n = 28) and intracapsular cataract extractions (ICCE) (n = 23). A final visual acuity was not available for 9 of the 60 patients.
727
Ophthalmology
Volume 100, Number 5, May 1993
Discussion The findings of this series contrast with the disease pattern of endophthalmitis caused by more virulent bacteria that presents characteristically over the first 72 hours. 8 The considerable variation in the delay from cataract surgery to diagnosis (range, 1 day to 28 weeks; median, 7 days; mean, 21 days) emphasizes the atypical pattern of endophthalmitis caused by the coagulase-negative staphylococci. There was a mean diagnostic interval of 7 days for the 22 patients with postsurgical coagulase-negative staphylococcal endophthalmitis described by Bode et al 9 and of 15 days for 13 such patients reported by Ficker et al.lO Forty-two percent of our patients presented more than 1 week, 22% more than 2 weeks, and 12% more than 1 month after surgery. We recorded a 29% prevalence of cataract wound abnormalities coincident with the diagnosis of endophthalmitis compared with a 22% prevalence reported by Driebe et al4 in a nonselected endophthalmitis series, although an actual wound leak was discerned in only 5% of our series, with 11 % similarly affected in Driebe et aI's patients. We do not believe that a poorly healed wound without aqueous humor drainage, in the presence of postoperative inflammation, implies that a wound leak has been the source of infection, although wound revision should be undertaken in such cases. Most of the delay in diagnosis was accounted for by a prolonged asymptomatic interval. 6,9 The variable progression of endophthalmitis in the interval between symptoms and diagnosis is also a significant observation. Over one quarter of the cataract surgery patients gave a history of symptoms lasting more than 72 hours before diagnosis and one tenth had symptoms for more than 7 days, and occasionally for much longer. Painless endophthalmitis II occurred in 10% of the cataract patients; this entity deserves much greater recognition. Overall, 49% of the cataract patients achieved visual acuity of ~20/50, compared with 85%, 55%, 38%, and 33% in the post-cataract coagulase-negative staphylococcal endophthalmitis cases ofO'Day et aI, 12 Bode et al,9 Ficker et aI, 13 and Bohigian and Olk, 14 respectively. Severity and disease variables differ among series, Nearly one third of our cataract patients with endophthalmitis retained less than 20/400 visual acuity. The often slowly progressive and delayed nature of the physical signs and symptoms emphasizes the need for immediate suspicion in patients with postoperative inflammation that is out of the ordinary. Improved familiarity with the spectrum of endo phthalmitis caused by these organisms (and other lesser pathogens) should enhance the management of delayedonset, subacute, chronic, and painless postoperative endophthalmitis. 5,9,10,12,13,15,16 Multiple regression models investigated the role ofnumerous iatrogenic and clinical factors (Table 2) in modeling the disease pattern, cognizant that bacterial virulence and host reactivity remained unmeasured. The clinical variation proved largely unrelated to details of the cataract surgery and postoperative management. The negative association between the asymptomatic period and topical
728
antibiotic prophylaxis suggests, perhaps, some role for topical antibiotic selection for more virulent bacteria. No predictive model could explain the often extended clinical course in these patients. In particular, neither the interval before symptoms developed nor the course of the symptoms was influenced significantly by major cataract surgery complications, the integrity of the lens capsule, the use of perioperative subconjunctival corticosteroids and/ or antibiotics, or the necessity for subsequent wound repair. These data support the premise that the use of perioperative corticosteroidsl 3 and antibiotics9 mask this disease, rather than delay its onset or alter fundamentally the disease process. We were unable to find evidence that the status of the posterior capsule was important in the disease spectrum. No significant differences existed in the clinical patterns between cases of ECCE or ICCE, with or without an IOL. It remains to be determined whether the enhanced adhesion of many coagulase-negative staphylococcal strains to the biomaterial I 7-20 (via extracellular bacterial biofilm 21 ,22) might be a pathogenetic factor in postoperative endophthalmitis. Painless endophthalmitis had no significant association with corticosteroid use. II ,23 The disease spectrum and outcome in coagulase-negative staphylococcal endophthalmitis are probably mainly due, therefore, to intrinsic differences in bacterial pathogenicity. The size and site of bacterial inoculation and varying bacterial virulence are likely to be important factors. 24,25 An intriguing and unexpected finding was the relatively strong direct association of a good visual outcome (protective effect) with the use of intensive topical corticosteroids in the symptomatic period before diagnosis. Paradoxically, the use of subconjunctival corticosteroids at cataract surgery, coincidental with the bacterial inoculation, was negatively correlated with outcome; this novel observation suggests that the suppression of early inflammatory events might modify adversely subsequent hostbacterial inflammatory and immunologic relationships in coagulase-negative staphylococcal endophthalmitis. Of particular interest was the failure to detect a significant association between visual outcome and such variables as the type of cataract surgery, major operative complications, the integrity of the lens capsule, subconjunctival antibiotic prophylaxis, endophthalmitis treatment with oral or intravitreal corticosteroids (confounded by associated vitrectomy and antibiotic administration), the asymptomatic and symptomatic periods, the retinal reflex at diagnosis, and the methicillin sensitivity of the coagulase-negative staphylococcal isolate. No inferences as to the effectiveness of treatment modalities could be made from these data because of management uniformity. This retrospective study suffers from the flaws inherent in any such investigation. However, our exploratory work approaches questions that have not been addressed previously, and multiple regression models provide a rigorous analysis of comprehensive data. The relative lack of influence of the multiplicity of peri operative variables in cataract surgery on the subsequent coagulase-negative en-
Ormerod et al . Postoperative Coagulase-negative Staphylococcal Endophthalmitis dophthalmitis suggests that, with the exception of corticosteroid use, they probably can be disregarded in the analysis of prospective studies of endophthalmitis management. Acknowledgments The authors thank the staff ophthal-
mologists at the Massachusetts Eye and Ear Infirmary, Boston, and the 39 cataract surgeons from across New England for their active contributions.
References 1. Kattan HM, flynn HW Jr, Pflugfelder SC, et al. Nosocomial endophthalmitis survey: current incidence of infection after intraocular surgery. Ophthalmology 1991 ;98:227-38. 2. Postoperative infections of the eye. In: Locatcher-Khorazo 0, Seegal BC, eds. Microbiology of the Eye. St. Louis: CV Mosby, 1972:77-85. 3. Rowsey JJ, Newsom DL, Sexton OJ, Harms WK. Endophthalmitis: current approaches. Ophthalmology 1982;89: 1055-66. 4. Driebe WT Jr, Mandelbaum S, Forster RK, et al. Pseudophakic endophthalmitis: diagnosis and management. Ophthalmology 1986;93:442-8. 5. Shrader SK, Band JD, Lauter CB, Murphy P. The clinical spectrum of endophthalmitis: incidence, predisposing factors, and features influencing outcome. J Infect Dis 1990;162:115-20. 6. Ormerod LD, Ho DO, Becker LE, et al. Endophthalmitis caused by the coagulase-negative staphylococci. 1. Disease spectrum and outcome. Ophthalmology 1993; 100:715-723. 7. Keppel G, Zedeck S. Data Analysis for Research Design: Analysis-of-Variance and Multiple Regression/Correlation Approaches. New York: WH Freeman, 1989. 8. Jaffe NS, Jaffe MS, Jaffe GF. Cataract Surgery and Its Complications, 5th ed. St Louis: CV Mosby, 1990:506-42. 9. Bode DO Jr, Gelender H, Forster RK. A retrospective review of endophthalmitis due to coagulase-negative staphylococci. Br J Ophthalmol 1985;69:915-9. 10. Ficker LA, Meredith TA, Wilson LA, Kaplan HJ. Role of vitrectomy in Staphylococcus epidermidis endophthalmitis. Br J Ophthalmol 1988;72:386-9. 11. Deutsch T A, Goldberg MF. Painless endophthalmitis after cataract surgery. Ophthalmic Surg 1984;15:837-40. 12. O'Day OM, Jones DB, Patrinely J, Elliott JH. Staphylococcus epidermidis endophthalmitis: visual outcome fol-
13. 14. 15. 16.
17.
18. 19.
20. 21. 22.
23. 24. 25.
lowing noninvasive therapy. Ophthalmology 1982;89:35460. Ficker L, Meredith TA, Wilson LA, et al. Chronic bacterial endophthalmitis. Am J Ophthalmol 1987;103:745-8. Bohigian GM, Olk RJ. Factors associated with a poor visual result in endophthalmitis. Am J Ophthalmol 1986;101:33241. Fox GM, Joondeph BC, flynn HW Jr, et al. Delayedonset pseudophakic endophthalmitis. Am J Ophthalmol 1991;111: 163-73. Weber OJ, Hoffman KL, Thoft RA, Baker AS. Endophthalmitis following intraocular lens implantation: report of 30 cases and review of the literature. Rev Infect Dis 1986;8: 12-20. Christensen GO, Simpson W A, Bisno AL, Beachey EH. Adherence of slime-producing strains of Staphylococcus epidermidis to smooth surfaces. Infect Immun 1982;37:31826. Hogt AH, Dankert J, de Vries JA, Feijen J. Adhesion of coagulase-negative staphylococci to biomaterials. J Gen Microbiol 1983; 129:2959-68. Herrmann M, Vaudaux PE, Pittet 0, et al. Fibronectin, fibrinogen, and laminin act as mediators of adherence of clinical staphylococcal isolates to foreign materials. J Infect Dis 1988;158:693-701. Griffiths PG, Elliott TSJ, McTaggart L. Adherence of Staphylococcus epidermidis to intraocular lenses. Br J Ophthalmol 1989;73:402-6. Costerton JW, Cheng K-J, Geesey GG, et al. Bacterial biofilms in nature and disease. Annu Rev Microbiol 1987;41: 435-64. Hartmann C, Jansen B, Schumacher-Perdreau E, Peters G. Bakterielle Spatendophthalmitis nach Kunstlinsenimplantation. Eine chronische, IOL-assoziierte "Plastikinfektion." In: Schott K, Jacobi KW, Freyler H, eds. 4. Kongrel3 der Deutschen Gesellschaft fUr Intraokularlinsen Implantation: 6. bis 7. April 1990, Essen, Berlin: Springer-Verlag, 1991: 170-8. Laatikainen L, Tarkkanen A. Early vitrectomy in the treatment of post-operative purulent endophthalmitis. Acta Ophthalmol 1987;65:455-60. Ormerod LD, Edelstein MAC, Schmidt GJ, et al. The intraocular environment and experimental anaerobic bacterial endophthalmitis. Arch Ophthalmol 1987;105:1571-5. Shockley RK, Jay WM, Fishman PH, et al. Effect ofinoculum size on the induction of endophthalmitis in aphakic rabbit eyes. Acta Ophthalmol 1985;63:35-8.
729