Endophthalmitis remains a rare but devastating complication of ocular surgery and trauma. In recent past several large series are published on postoperative12345678910111213 and posttrauamtic endophthalmitis1415161718 (Pubmed search using the key words: endophthalmitis, postoperative, trauma.) One large multicentre, prospective study, the Endophthalmitis Vitrectomy Study (EVS),19 has described the relationship between clinical presentation and clinical outcome in postcataract/secondary intraocular lens (IOL) acute bacterial endophthalmitis. The EVS protocol denied recruitment of severe cases of endophthalmitis, fungal endophthalmitis and posttraumatic endophthalmitis patients.
We herein report a large single-centre, prospective study describing the relationship between clinical presentation and visual outcome. This may be of interest for two important reasons: 1) to validate the findings of the other studies on this topic; and 2) to provide single-centre information, which ensures greater uniformity in data. In this study, we prospectively analysed 206 and 182 patients with postoperative and posttraumatic endophthalmitis respectively and performed a multivariate analysis of presenting clinical features to determine independent risk factors for poor visual outcome in each instance.
Materials and Methods
These 388 patients were seen between 1991-1997 at the L V Prasad Eye Institute, a large tertiary care eye center in Hyderabad, south central India. Of these, 128 and 108 patients of postoperative and posttraumatic endophthalmitis respectively were followed for a minimum of 3 months. Following the Institutional Review Board approval a database was expressly designed to collect clinical information on endophthalmitis patients. All patients were examined by one of the three vitreoretinal surgeons (TD, SJ, and ABM). The detailed protocol included demographic information (gender and age), history (predisposing ocular surgery/ trauma, complications of the surgery, duration of symptoms, interval between the event and presentation to this institution, and use of medications (topical or systemic), examination findings (presenting visual acuity, presence of hypopyon, presence of vitreous cells, and optic disc visibility), ultrasonography (with attention to the presence of vitreous membranes), and microbiology (including culture and susceptibilities).
All patients in this study received vitrectomy. Eyes were prepared with 5% povidone-iodine solution. After sclerotomy, but before initiating the infusion fluid, 0.5-0.75 ml of undiluted vitreous was manually aspirated from mid-vitreous into a 3-ml syringe for microbiology (microscopy and culture). Once the sample was obtained, the infusion was initiated and automated cutting- suction continued with collection into a vitrectomy cassette. It was the goal of surgery to remove the core vitreous with an attempt to clear as much vitreous debris as possible. Surgery was considered adequate if visualisation of the disc and second retinal vessels was possible.
Using standard techniques,20 the undiluted vitreous biopsy samples were immediately inoculated directly onto sheep blood agar, chocolate agar, non-nutrient agar, Sabouraud′s dextrose agar, potato dextrose agar, thioglycolate, and brain heart infusion broth. Sabouraud′s and potato dextrose agar plates were incubated at 250C to enhance the growth of fungi, and the remainder was incubated at 370C. Blood agar plates were incubated under aerobic and anaerobic conditions, and chocolate agar was incubated with 5% carbon dioxide. Gram stain, Giemsa stain, and KOH with calcofluor white under fluorescence were included as part of the standard protocol for microscopic evaluation of biopsy samples. A culture was considered positive when there was growth of the same organism on two or more media, confluent growth at site of inoculation on one solid medium, or growth in one medium with consistent direct microscopy findings.
All patients received intravitreal and topical antibiotics, and those not suspected to have fungal endophthalmitis also received oral corticosteroids. Intravitreal antibiotics included amikacin (0.4 mg in 0.1 ml) or ceftazidinme (2.25 mg in 0.1 ml) for empiric gram-negative organism coverage and vancomycin (1 mg in 0.1 ml) or cefazoline (2.25 mgin 0.1 ml) for empiric gram-positive organism coverage. The topical antibiotics included the fortified gentamicin (40 mg/ml), cefazoline (50 mg/ml), and/or amphotericin B. Approximately half the patients received intravenous antibiotics, according to surgeon preference, which included gentamicin (4 mg/kg in divided doses) and cefazoline (40 mg/kg in divided doses). Patients receiving aminoglycosides underwent renal function screening (serum BUN and creatinine) prior to surgery and postoperative day 5. Patients receiving ketoconazole (200 mg twice daily) underwent liver function tests before surgery, and then once a fortnight until therapy was discontinued.
Statistical analysis was done as follows. Frequencies and percentages were reported for all variables. As the objective of the study was to determine presenting factors associated with final vision, visual acuity at 3 months follow-up visit was classified for the purpose of analysis into 2 variables, namely worse than (<) 6/120 and worse than (<) 6/18. All factors believed to be associated with final visual acuity were initially analysed using the chi square univariate test. Following this, multiple logistic regression analysis was used to determine the strength of association of each variable after adjusting for the rest of the variables. Only those variables that were significant in the univariate analysis, and were felt to be clinically reasonable, or were independent of all other variables were included into the logistic model. The enter method was used to determine the effect of each factor in the presence of all other factors. The order of importance of each independent significant factor in the model was determined using the forward stepwise method of logistic regression. Odds ratios and 95% confidence limits were reported. SPSS (Statistical Package for the Social Science), version 8.0 was used for analysis.
Postoperative endophthalmitis. Baseline characteristics of the study population are listed in Table 1. The median age was 59 (range 3-85) years, with 135 (65.56%) males and 107 (51.9%) left eyes involved. Predisposing surgeries included cataract surgery (159 patients, 77%), glaucoma filtration surgery (6 patients, 3%), vitrectomy (8 patients, 4%), and YAG capsulotomy for post-cataract surgery- posterior cortical opacification (6 patients, 3%). Complications of the primary surgery included posterior capsule rupture and vitreous loss, each present in 10% (20 of 206) of the study population. The median duration of symptoms was 4 days, and the median interval between predisposing surgery and presentation to this institution was 6 days. The presenting visual acuity was light perception (LP) or worse in 128 (62.1%) patients, detection of hand motions (HM) in 59 (28.6%) and 6/120 or better in 19 (9.2%) patients. Clinical examination showed hypopyon in 157 (76.2%) and vitreous cells in 122 (59.2%) eyes. The optic disc was not visible by indirect ophthalmoscopy in 167 (81.1%) eyes. Vitreous membranes were seen in 118 (57.3%) patients on ultrasonography. Vitreous biopsy was culture positive in 113 (54.9%), with bacteria cultured in 99 (48.1%) and fungus cultured in 20 (9.7%) of cases (sum is greater than total of 55% because of polymicrobial cultures). The microbial spectrum of postoperative endophthalmitis from this institution has been reported previously.21
At 3 months, 25 (19.5%) of 128 patients regained a visual acuity of 6/12 or better, 34 (26.6%) patients 6/18 or better, 59 (46.1%) patients 6/36 or better, 75 (58.6%) patients 6/60 or better, and 78 (60.9%) patients 6/120 or better Figure 1. Thirty-one (24.2%) of 128 patients had a visual acuity of LP or worse, and of these 9 (7.0%) were no light perception (NLP). Ten patients did not complete 3 months′ follow-up though they were doing well at the one-month follow-up visit.
The number of patients included in the multivariate analysis is listed in Table 1, so also the number not included in the multivariate analysis (due to lack of visual acuity data at 3 months, the endpoint of this study). Chi-square for proportions analysis showed that the two groups were similar at baseline, with the exception of a higher proportion of males in the non-analysed group (P=0.017).
Table 2 shows the results of the univariate analysis for determining clinical presentation factors associated with visual outcome, divided into two groups, namely < 6/18 and < 6/120. The presenting features associated with < 6/18 visual acuity at 3 months included a history of intracapsular cataract (ICCE) surgery (P=0.032), poor visual acuity at presentation (P< 0.0001), the presence of vitreous cells (P=0.023), inability to visualise the optic disc on indirect ophthalmoscopy (P=0.001), and the presence of vitreous membranes on ultrasound (P=0.05). Non-significant features are also listed (Table 2). The factors associated with < 6/120 at 3 months included a history of surgery other than extra capsular cataract extraction (ECCE) (P=0.052), poor visual acuity at presentation (P=0.001), inability to visualise the disc on indirect ophthalmoscopy (P=0.019), presence of vitreous membranes on ultrasound (P=0.001), any organism cultured from vitreous biopsy (P=0.017), bacteria cultured from vitreous biopsy (P=0.008), and gram-negative bacteria cultured from vitreous biopsy (P=0.02).
Table 3 shows the results of the multivariate analysis for determining independent risk factors for poor visual outcome in postoperative endophthalmitis. Factors included in the multivariate analysis are indicated by an asterisk in Table 2. Analysis revealed that poor visual acuity at presentation was the only independent risk factor for visual acuity < 6/18 at 3 months follow-up. Further it suggested that a patient presenting with - LP is about 6-times more likely to have a final visual acuity of < 6/18, and patients with vitreous membranes on ultrasonography at presentation was about 2.5-times more likely to have a final visual acuity of < 6/120. Two features found significant in the univariate analysis were excluded from the multivariate analysis; these were, predisposing surgery other than ECCE (excluded for lack of clinical relevance) and any bacteria cultured from vitreous biopsy (excluded because of clinical redundancy with other microbiology factors).
Figure 1 shows the trend of visual recovery at 3 months follow-up in postoperative endophthalmitis.
Posttrauamtic endophthalmitis. Baseline characteristics of the study population are listed in Table 4. The median age was 16 (range 1-72) years, with 140 (76.9%) males and 98 (53.8%) right eyes involved. Mechanism of injury included injury by arrow (n=7 patients; 4%), glass (n=3 patients; 2%), insect (n=2 patients; 1%), nail (n=4 patients; 2%), needle (n=30 patients; 17%), stone (n=15 patients; 8%), thorn (n=21 patients; 12%), and wood (n= 40 patients; 22%). Retained intraocular foreign body (IOFB) was present in 35 (19%) patients. The median duration of symptoms was 5 days. The presenting visual acuity was ≤ LP in 132 (72.5%) patients, HM in 28 (15.4%), and ≥ 6/120 in 19 (10.4%) patients. Clinical examination showed hypopyon in 131 (72%) patients, vitreous cells in 74 (40.7%), corneal laceration in 135 (74.2%), and inability to see the optic disc in 152 (83.5%) patients. The vitreous membranes were seen on ultrasonography in 112 (61.5%) of the study population. The vitreous culture was positive in 118 (64.8%), with bacteria cultured in 104 (57.1%) and fungus cultured in 19 (10.4%) cases (sum is greater than total of 65% because of polymicrobial cultures). The microbial spectrum of posttraumatic endophthalmitis from this institution has been reported previously.22
At 3 months, 32 (29.6%) of 108 patients regained 6/12 or better, 41 (38.0%) 6/18 or better, 58 (53.7%) 6/36 or better, 66 (61.1%) 6/60 or better, and 69 (63.9%) 6/120 or better (Figure 1). Thirty-one (28.7%) of 108 patients had a visual acuity of LP or worse and of these 8 (7.4%) were NLP at 3-months′ follow-up.
The number of patients included in the multivariate analysis is listed in Table 4, so also the number not included in the multivariate analysis (due to lack of visual acuity data at three months, the endpoint of this study). Chi-square for proportions analysis suggested that the two groups were similar at baseline, with the exception of younger median age in the group lost to follow-up (non-analysed group), as well as a higher proportion of retained IOFB (P=0.006) and a lower proportion of vitreous cells present (P=0.003) in the analysed group.
Table 5 shows the results of the univariate analysis for determining clinical presentation factors associated with visual outcome, divided into two groups, namely < 6/18 and < 6/120. The features associated with < 6/18 at 3 months included retained IOFB (P=0.011), trauma by needle (P=0.044), trauma by mechanism other than thorn (P< 0.0001), poor visual acuity at presentation (P=0.003), inability to visualize the optic disc on indirect ophthalmoscopy (P=0.024), and the presence of vitreous membranes on ultrasound (P=0.014). Nonsignificant features are also listed (Table 5). The factors associated with < 6/120 at 3 months included a retained IOFB (P=0.025), a history of trauma by needle (P=0.008), trauma by mechanism other than wood (P=0.038), duration of signs/symptoms < 6 days (P=0.034), inability to visualise the optic disc on indirect ophthalmoscopy (P=0.038), and a culture positive for any organism (P=0.003).
Table 6 shows the results of the multivariate analysis for determining independent risk factors for poor visual outcome. Factors included in the multivariate analysis are indicated by an asterisk in Table 5. Analysis demonstrated that retained IOFB was an independent risk factor for visual acuity < 6/18 at 3 months′ follow-up. It suggested that a patient presenting with a retained IOFB was about 6-times more likely to have a final visual acuity of < 6/18. It further revealed that trauma by needle and retained IOFB were independent risk factors for visual acuity of < 6/120 at 3 months′ follow-up. A patient with a history of trauma by needle or a retained IOFB was about five/four times more likely, respectively, to have a final visual acuity of < 6/120. Three features found significant in the univariate analysis were excluded from the multivariate analysis because of correlation to another variable or lack of clinical relevance; these were, (1) trauma by mechanism other than thorn, (excluded for lack of clinical relevance), (2) trauma by mechanism other than wood (excluded for lack of clinical relevance), and (3) vitreous membranes seen on ultrasonography (excluded because it correlated with inability to visualise the optic disc). Figure shows the trend of visual recovery at 3 months followup in posttraumatic endophthalmitis.
Endophthalmitis is a dreaded complication of both intraocular surgery and trauma. In an attempt to suitably manage this disease, numerous authors have examined endophthalmitis by infecting organism in both postoperative and posttraumatic cases,232425262728 and the mechanism of injury in posttraumatic endophthalmitis cases.14151618 The main objective of this study was to determine whether any clinical presentation features were associated with poor visual outcome. Retrospective studies on postoperative2930 and posttraumatic endophthalmitis1415 have reported that poor visual recovery was associated with a poor presenting vision, larger wound length (pertaining to trauma), a positive culture, more virulent organisms, a delay in diagnosis and treatment, compromised immunological status, and concomitant ocular disease such as rubeosis iridis, periretinal proliferation, vitreous haemorrhage, and retinal detachment. The EVS reported that older age, a history of diabetes, a corneal infiltrate or ring ulcer, abnormal intraocular pressure, rubeosis iridis, an absent red reflex, an afferent pupillary defect, an open posterior capsule, and visual acuity of LP only were independent risk factors for decreased visual acuity.19
For the purpose of analysis we had two groups, 6/18 and 6/120. The former represented the least reading vision with appropriate aids and the later represented the legal blindness. In the postoperative endophthalmitis patients in this series, a 3-month postoperative visual acuity of < 6/18 was associated with 5 factors; they were a history of ICCE, poor initial visual acuity, vitreous cells, a non-visible optic disc, and vitreous membranes seen on ultrasonography. Visual acuity of < 6/120 was associated with seven factors; they were a history of surgery other than ECCE, poor initial visual acuity, a non-visible optic disc, vitreous membranes on ultrasonography, a culture positive for any organism, a culture positive for any bacteria, and a culture positive for gram-negative bacteria. In the posttraumatic endophthalmitis in this series 3-month post treatment visual acuity of < 6/18 was associated with 5 clinically relevant factors; they were a retained IOFB, trauma by needle, poor visual acuity at presentation, inability to visualise the optic disc, and the presence of vitreous membranes on ultrasonography. Visual acuity of < 6/120 was associated with five clinically relevant factors; they were a retained IOFB, trauma by needle, duration of sign/symptoms ≤ 6 days, inability to visualise the optic disc, and a culture positive for any organism.
In either situation associations that lacked clinical relevance (as in case of vitreous cells, excluded in favor of optic disc not visualised) and overlapping association (as in case of microbiology factors) were removed from the multivariate logistic regression (Table 2 and 4).
In postoperative endophthalmitis the visual acuity of ≤ LP at presentation was an independent risk factors for a 3-month postoperative visual acuity of < 6/18, with an odds ratio of 5.85 [1.25-27.42, 95% CI]. This is similar to the EVS data which found that a patient presenting with a visual acuity of LP is twice as likely to have decreased visual acuity compared to one presenting with visual acuity of > LP.19
In posttraumatic endophthalmitis a retained IOFB was an independent risk factor for a 3-month post treatment visual acuity of < 6/18, with an odds ratio of 5.9 [1.85 - 18.75, 95% CI]. A history of trauma by needle and a retained IOFB were independent risk factors for a 3-month posttreatment visual acuity of < 6/120, with odds ratios of 4.47 [1.22 - 16.38, 95% CI] and 3.76 [1.36 - 10.37, 95% CI], respectively.
In both postoperative and posttraumatic endophthalmitis one quarter to one third patients attained 6/18 or better at 3 months and nearly 60% patients regained at least 6/60. Although posttraumatic and postoperative endophthalmitis are separate clinical entities, it may be of interest to compare presentation characteristics of the two populations (Tables 1 and 4). Notable differences between the two populations validate well-known clinical findings. The median age for posttraumatic endophthalmitis was 16 years, versus 59 years for postoperative endophthalmitis. Likewise, posttraumatic endophthalmitis was much more likely to be caused by Bacillus species than postoperative endophthalmitis (11.0% versus 1.0%, p=0.0005).
Limitations of this study include those fundamental to statistical studies. First, statistical studies of this size can result in significant associations that make little clinical sense. These associations were excluded from logistic regression analysis. Second, the analysis in this study did not allow us to combine variables. For instance, a poor presenting visual acuity with a culture-positive vitreous biopsy may be a strong predictor of poor visual outcome though, only single variables were analysed in this study. It may also appear that combining different postoperative endophthalmitis aetiologies (such as acute postoperative endophthalmitis from cataract surgery and delayed-onset endophthalmitis from YAG capsulotomy) might skew results. But the multivariate logistic regression accounted for different dependent variables (such as operative aetiology) and allows us to parse variables and analyze multiple patient characteristics without skewing results. The other limitation of the study could be the analysis of the last decade data, 1991-1997. We have published the microbiology data of this period earlier. We believe that this analysis of clinical features and the risk factor analysis will complement the other reports. Further, this data will help us compare the more recent data (1998 onwards) and look for any alterations in the trends and outcome. Two Indian single centre studies on postoperative endophthalmitis have been published recently.1213 The similarity was the greater gram negative organism in all studies, but none have done the risk factors analysis. Our study is likely to bridge this gap.
In summary, this is the largest, single-centre, prospective study on risk factors for poor visual outcome in postoperative and posttraumatic endophthalmitis. It is hoped that recognition of the risk factors identified in this study will aid clinicians in discussing the visual prognosis in their endophthalmitis patients.
1. Fisch A, Salvanet A, Prazuck T, Forestier F, Gerbaud L, Coscas G, et al Epidemiology of infective endophthalmitis in France Lancet. 1991;338:1373–76
2. Kattan HM, Flynn HW, Pflugfelder SC, Robertson C, Forster RK. Nosocomial endophthalmitis survey: current incidence of infection after intraocular surgery Ophthalmology. 1991;98:227–38
3. Javitt JC, Vitale S, Canner JK, Street DA, Krakauer H, McBean M, et al National outcomes of cataract extraction. Endophthalmitis following in-patient surgery Arch Ophhtalmol. 1991;109:1085–89
4. Javitt JC, Street DA, Tielsch JM, Wang Q, Kolb MM, Schein O, et al National outcomes of cataract extraction. Retinal detachment and endophthalmitis following out-patient surgery Ophthalmology. 1994;101:100–5
5. Eifrig CW, Flynn HW Jr, Scott IU, Newton J. Acute-onset postoperative endophthalmitis: review of incidence and visual outcomes (1995-2001) Opthalmic Surg Lasers. 2002;33:373–78
6. Kamalarajah S, Silvestri G, Sharma N, Khan A, Foot B, Ling R, et al Surveillance of endophthalmitis following cataract surgery in the UK Eye. 2004;18:555–56
7. Wong TY, Chee SP. The epidemiology of acute endophthalmitis after cataract surgery in an Asian population Ophthalmology. 2004;111:699–705
8. Sandvig KU, Dannevig L. Postoperative endophthalmitis: establishment and result of a national registry Cataract Refract Surg. 2003;29:1273–80
9. Morlet N, Semmens LJ, Ng J, team EPSWA. The endophthalmiits population study of Western Australia (EPSWA): First report Br J Ophthalmol. 2003;87:574–76
10. Schmitz S, Dick HB, Krummenauer F, Pfeiffer N. Endophthalmitis in cataract surgery: Results of a German survey Ophthalmology. 1999;106:1869–77
11. Versteegh MF, Van Rij G. Incidence of endophthalmitis after cataract surgery in the Netherlands: Several surgical techniques compared Doc Ophthalmol. 2000;100:1–6
12. Gupta A, Gupta V, Gupta A, Dogra MR, Pandav SS, Ray P, et al Spectrum and clinical profile of post cataract surgery endophthalmitis in North India Indian J Ophthalmol. 2003;51:139–45
13. Anand AR, Therese L, Madhavan HN. Spectrum of aetiological agents of postoperative endophthalmitis and antibiotic susceptibility of bacterial isolates Indian J Ophthalmol. 2000;48:123–28
14. Thompson WS, Rubsamen PE, Flynn HW, Schiffman J, Cousins SW. Endophthalmitis after penetrating trauma: risk factors
and visual acuity outcomes Ophthalmology. 1995;102:1696–701
15. Williams DF, Mieler WF, Abrams GW, Lewis H. Results and prognostic factors in penetrating ocular injuries with retained intraocular foreign bodies Ophthalmology. 1988;95:911–16
16. Thompson JT, Parver LM, Enger CL, Mieler WF, Liggett PE. Infectious endophthalmitis after penetrating injuries with retained intraocular foreign bodies Ophthalmology. 1993;100:1468–74
17. Alfaro DV, Roth D, Liggett PE. Post-traumatic endophthalmitis: Causative organisms, treatment, and prevention Retina. 1994;14:206–11
18. Jalali S, Das T, Majji AB. Hypodermic needles: a new source of penetrating ocular trauma in Indian children Retina. 1999;19:213–17
19. Endophthalmitis Vitrectomy Study Group. . Results of the Endophthalmitis Vitrectomy Study: a randomized trial of immediate vitrectomy and of intravenous antibiotics for the treatment of postoperative bacterial endophthalmitis Arch Ophthalmol. 1995;113:1479–96
20. Sharma S, Jalali S, Adiraju MV, Gopinathan U, Das T. Sensitivity and predictability of vitreous cytology, biopsy, and membrane filter culture in endophthalmitis Retina. 1996;16:525–29
21. Kunimoto DY, Das T, Sharma S, Jalali S, Majji AB, Gopinathan U, et al Microbiologic spectrum and susceptibility of isolates: Part I. Postoperative endophthalmitis Am J Ophthalmol. 1999;128:240–42
22. Kunimoto DY, Das T, Sharma S, Jalali S, Majji AB, Gopinathan U, et al Microbiologic spectrum and susceptibility of isolates: Part II. Posttraumatic endophthalmitis Am J Ophthalmol. 1999;128:242–44
23. Mao LK, Flynn HW, Miller D, Pflugfelder SC. Endophthalmitis caused by Staphylococcus aureus Am J Ophthalmol. 1993;116:584–89
24. Irvine WD, Flynn HW, Miller D, Pflugfelder SC. Endophthalmitis caused by gram-negative organisms Arch Ophthalmol. 1992;110:1450–54
25. Mao LK, Flynn HW, Millder D, Pflugfelder SC. Endophthalmitis caused by streptococcal species Arch Ophthalmol. 1992;110:798–801
26. Endophthalmitis Vitrectomy Study Group. . Microbiologic factors and visual outcome in the Endophthalmitis Vitrectomy Study Am J Ophthalmol. 1996;122:830–46
27. Schemmer GB, Driebe WT. Posttraumatic Bacillus cereus endophthalmitis Arch Ophthalmol. 1987;105:342–44
28. Das T, Chowdhury K, Sharma S, Jalali S, Nuthethi Rthe Endophhtlmitis research Group. . Clinical profile and outcome in Bacillus endophthalmitis Ophthalmology. 2001;108:1819–25
29. Bartz-Schmidt KU, Bermig J, Kirchhof B, Wiedemann P, Walter P, Heimann K. Prognostic factors associated with the visual outcome after vitrectomy for endophthalmitis Graefe′s Arch Clin Expt Ophthalmol. 1996;234:S51–58
30. Bohigian GM, Olk RJ. Factors associated with a poor visual results in endophthalmitis Am J Ophthalmol. 1986;101:332–34