Purpose
To determine risk factors for poor visual acuity outcomes in patients with endogenous Klebsiella pneumoniae endophthalmitis.
Design
Retrospective case-control study.
Methods
We reviewed all consecutive patients diagnosed with endogenous Klebsiella pneumoniae endophthalmitis over 20 years. Data collected included patients’ demographics, clinical findings, investigations, and therapeutic intervention. Main outcome measures were visual acuity (VA) at 3 months postinfection and/or evisceration. Poor vision was defined as VA worse than 4/200.
Results
Seventy-one eyes of 61 patients were studied. Mean age was 55.7 (standard deviation [SD] ± 13.1) years. Majority were male (49/61, 80.3%), Chinese (51/61, 83.6%), and had hepatobiliary sepsis (47/61, 77.5%). Mean time to ocular symptoms was 4.4 (± 3.0) days. Fifty-four of 71 eyes (80.2%) had poor vision and 19/71 eyes (26.8%) required evisceration. The most significant risk factor was hypopyon (41/71 eyes, 57.7%) on multivariate analysis (odds ratio [OR], 52.6; 95% confidence interval [CI], 1.7–1000; P = .01). Unilateral involvement (OR, 10.4; 95% CI, 1.2–90.2; P = .01) and patients that were managed before year 2000 (before routine screening was implemented) (OR, 5.2; 95% CI, 1.5–17.9; P = .037) were significant risk factors for evisceration on multivariate analysis.
Conclusion
Patients presenting with hypopyon and unilateral involvement have a poorer prognosis.
Endogenous Klebsiella pneumoniae endophthalmitis is known to be most prevalent in East Asia, with an increasing number of cases being reported in Asia and around the world. It is a rare but devastating complication of Klebsiella pneumoniae bacteremia, where despite early diagnosis and treatment, the visual prognosis remains dismal, with a high likelihood of bilateral blindness. The most common source of infection is hepatobiliary sepsis, with a reported incidence of 3.0% to 7.8% in patients with K. pneumoniae liver abscess. The predisposition of East Asians to Klebsiella hepatobiliary sepsis is uncertain. However, with the advent of globalization and immigration, K. pneumoniae has overtaken Escherichia coli as the leading cause of liver abscesses in the United States. Even with the administration of appropriate antibiotic therapy, and percutaneous surgical drainage of the abscess if necessary, this severe disease has a significant mortality rate (6%-14%).
A review of current literature shows that endogenous Klebsiella pneumoniae endophthalmitis has been discussed in separate case reports and several small series. Data from these studies suggest techniques and outcomes of early intervention and evaluate risk factors in patients with K. pneumoniae sepsis for developing endogenous Klebsiella pneumoniae endophthalmitis. However, we are unable to find studies that evaluate predictors for poor visual outcome and evisceration. The aim of our study is to identify risk factors for poor visual outcomes in endogenous Klebsiella pneumoniae endophthalmitis.
Methods
We identified all cases of endogenous bacterial endophthalmitis seen at the Singapore National Eye Center or the affiliated Singapore General Hospital between January 1, 1986 and January 1, 2007 and Changi General Hospital from January 1, 2000 to January 1, 2007. Patients were diagnosed as having endogenous Klebsiella pneumoniae endophthalmitis and included in the study if they had 1) clinical signs consistent with endogenous endophthalmitis such as the presence of severe anterior chamber inflammation, vitreous haze, or choroidal abscess, 2) absence of any other causes of intraocular inflammation, and 3) K. pneumoniae isolated from blood, intraocular fluid, or any other tissue sample.
Patients with endogenous Klebsiella pneumoniae endophthalmitis either presented directly to the ophthalmologist with eye symptoms or were referred from the medical or surgical unit of Singapore General Hospital or Changi General Hospital, to which they had been admitted for sepsis with or without ocular symptoms. After the year 2000, because of the increased awareness among physicians of endogenous Klebsiella pneumoniae endophthalmitis as an important complication of K. pneumoniae sepsis, patients admitted with sepsis were routinely referred to the ophthalmology service for screening. All patients were managed by one primary ophthalmologist in each center (S.P.C. and A.J.) and co-managed with the infectious diseases physicians. Visual acuity (VA) was taken in the ophthalmology clinic and checked with a Snellen chart. Patients suspected of endogenous Klebsiella pneumoniae endophthalmitis were investigated with a standard protocol, which included complete blood count, erythrocyte sedimentation rate, liver enzyme panel, blood and urine microscopy and culture, imaging such as chest radiograph and ultrasonography of hepatobiliary system, and aqueous or vitreous microscopy and culture. After vitreous sampling, patients were given intravitreal vancomycin (2.0 mg/0.1 mL) plus ceftazidime (2.25 mg/0.1 mL) or amikacin (0.4 mg/0.1 mL). Patients who presented directly to ophthalmologists were started on intravenous (IV) ceftriaxone as first-line therapy; those who were referred from the medical or surgical unit were continued on the pre-existing IV antibiotics. This was then changed according to the culture and sensitivity results, if there was no clinical improvement. Intravitreal antibiotics were repeated after 48 to 72 hours in severe cases. Since 1995, trans–pars plana vitrectomy (TPPV) was performed if the eye condition continued to deteriorate or developed a worsening grade of relative afferent pupillary defect, or if there was no response to intravitreal and IV antibiotics within 48 hours. Evisceration was performed if perforation occurred or if there was progressive panophthalmitits, or in painful blind eyes.
Patient demographics, ocular history, and systemic and ocular examination findings were collected. The time interval from the onset of sepsis to onset of ocular symptoms, symptoms at presentation, and significant comorbidities were recorded. Clinical features recorded included presenting VA and VA at 3 months postinfection, ocular findings, presence of hypopyon, and intraocular pressures. VA was analyzed as a categorical variable, with 2 subgroups: visual acuity worse than 4/200 and visual acuity 4/200 and better. Investigations such as the source of infection; microbiologic culture results from blood, vitreous, or other tissue samples; and interventions such as systemic antibiotics, intravitreal antibiotics, and vitrectomy were also recorded. Cases were classified according to the recommendation by Greenwald and associates. A poor visual outcome at 3 months post resolution of infection was defined as VA worse than 4/200 and/ or evisceration, similar to previous reviews on endogenous bacterial endophthalmitis. The minimum follow-up was 3 months, as VA at 3 months was 1 of our outcome measures unless evisceration was required at an earlier time interval. Prognostic factors for poor visual outcome were analyzed, including laterality, classification of endophthalmitis according to location, time interval to ocular symptoms, presence of hypopyon, increased intraocular pressure, presenting VA, diabetes mellitus, use of intravitreal antibiotics, and whether vitrectomy was performed.
Statistical significance was set at P < .05 and SPSS version 17.0 (SPSS Inc., Chicago, Illinois, USA) was used for the analysis. Age- and gender-adjusted multivariate logistic regression was used to determine significant risk factors and the odds ratios (OR) with 95% confidence intervals (CI) were calculated.
Results
Of the 113 patients with culture-proven endogenous bacterial endophthalmitis seen at both centers over the study period, 71 eyes of 61 patients had culture-proven endogenous Klebsiella pneumoniae endophthalmitis, and were included in our study. The patient characteristics, ocular features, microbiology results, and outcomes are summarized in Table 1 .
| Demographics | |
| Patients | |
| Number | 61 |
| Eyes affected | 71 |
| Mean age, years (± SD) | 55.7 (±13.1) |
| Sex | |
| Male | 49 (80.3%) |
| Female | 12 (19.7%) |
| Race | |
| Chinese | 51 (83.6%) |
| Non-Chinese (Malay, Indian) | 10 (16.4%) |
| Systemic illness | |
| Diabetes mellitus | 34 (55.7%) |
| Immunosuppression | 2 (3.3%) |
| Source of sepsis | |
| Hepatobiliary tract | 55 (77.5%) |
| Urinary tract | 7 (9.9%) |
| Respiratory | 3 (4.2%) |
| Other | 6 (8.4%) |
| Clinical features | |
| Presenting visual acuity | |
| Poor: VA worse than 4/200 | 62 (87.3) |
| Good: 20/400 or better | 9 (12.6%) |
| Type | |
| Panophthalmic | 28 (39.4%) |
| Posterior diffuse | 33 (46.5%) |
| Posterior focal | 9 (12.7%) |
| Anterior focal | 1 (1.4%) |
| Hypopyon | |
| Yes | 51 (71.8%) |
| No | 20 (28.2%) |
| Laterality | |
| Unilateral | 51 (83.6%) |
| Mean time to ocular symptoms, days (± SD) | 4.40 ( ± 5.00) |
| Vitrectomy | 19 (26.8%) |
| Outcomes | |
| Visual acuity 3 months postinfection | |
| Poor: VA worse than 4/200 | 55 (77.5%) |
| Good: 20/400 or better | 16 (22.5%) |
| Evisceration | 19 (26.8%) |
Patient Characteristics
The mean age of the endogenous Klebsiella pneumoniae endophthalmitis patients was 55.7 (standard deviation [SD] ± 13.1) years. They were predominantly male (n = 49/61, 80.3%) and Chinese (n = 51/61, 71.8%). The most common underlying medical condition was diabetes mellitus (n = 34/51, 55.7%). Twenty-five of 61 patients (41.0%) had no underlying illness and were not immunocompromised. The time interval from onset of sepsis to ocular symptoms ranged from a simultaneous onset to 30 days. The mean time interval was 4.4 (± 3.0) days. In 3 patients (4.9%), ocular presentation was the initial feature of sepsis. These 3 patients had none or minimal systemic symptoms with no underlying predisposing factors, and the K. pneumoniae sepsis were diagnosed only following investigations.
Ocular Features
Common presenting complaints were ocular pain, blurring of vision, and eye redness. Presence of hypopyon was found in 49.3% (35/71 eyes) while 16 eyes had a characteristic “pupillary hypopyon” with fibrinous exudates obscuring the pupil. Increased intraocular pressure was also found in 11 of 71 eyes (15.4%) at time of presentation. The majority of the patients (n = 51/61, 83.6%) had unilateral involvement. Most patients had poor VA at presentation, with 87.3% (62/71 eyes) having VA worse than 4/200. The most common type of endophthalmitis was posterior diffuse involvement (33/71 eyes, 46.5%), followed by panophthalmic involvement (28/71 eyes, 39.4%) and posterior focal involvement (9/71 eyes, 12.7%). There was 1 case of anterior focal endophthalmitis.
Microbiology
The commonest source of infection was from the hepatobiliary system (55/71 eyes, 77.5%), followed by the urinary tract (7/71 eyes, 9.9%). In our series of culture-proven endogenous Klebsiella pneumoniae endophthalmitis patients, 31 of 71 eyes (43.6%) were positive from intraocular samples, while 92.9% (66/71 eyes) were blood culture positive and 5 cases culture positive from other sources (eg, pulmonary, cerebrospinal fluid). Twenty-eight of 71 eyes (39.4%) had both ocular and blood cultures that were positive. Patients who had been admitted initially were already started on a broad-spectrum antibiotic (IV ceftriaxone) at the time of the ophthalmology referral.
Outcomes and Prognostic Factors
Sixty of 71 eyes (84.5%) received intravitreal antibiotics. Nineteen of 71 eyes (26.8%) underwent vitrectomy as indicated. The visual outcome was generally unfavorable, with 54 of 71 eyes (76.0%) having poor visual outcomes. A total of 57.8% (41/71 eyes) had a VA of no light perception. Nineteen eyes (26.8%) eventually required evisceration to eradicate infection following loss of vision, or to remove phthisical eyes ( Table 1 ). None of the patients received intravitreal corticosteroids at the first instance, and only 13 patients received corticosteroids on the second or third intravitreal tap.
The prognostic factors for poor visual outcome and evisceration are summarized in Table 2 . Age- and gender-adjusted logistic regression revealed that the presence of hypopyon (OR, 19.2; 95% CI, 3.2–111; P = .001), unilateral involvement (OR, 5.5; 95% CI, 1.6–19.2; P = .008), time interval from onset of sepsis to ocular symptoms (OR, 0.8; 95% CI, 0.7–1.0; P = .015), and panophthalmic involvement (OR, 23.8; 95% CI, 3.0–200; P = .003) were risk factors for poor visual outcome. The patients who developed ocular involvement rapidly (less than 4 days from onset of sepsis) were more likely to have panophthalmic involvement and hypopyon (OR, 1.5; 95% CI, 0.6–3.9; P = .03), as compared to those who had a more indolent course (more than 4 days).
| Age- and Gender-Adjusted a | Multivariate a | |||||
|---|---|---|---|---|---|---|
| OR (95% CI) | P Value | OR (95% CI) | P Value | |||
| Risk factors for poor visual outcome | ||||||
| Unilateral involvement | Y = 51 | n = 44 (86.3) | 5.5 (1.6–19.2) | .008 | 12.05 (0.985–142.86) | .051 |
| N = 20 | n = 11 (55.0) | |||||
| Panophthalmic involvement b | Y = 27 | n = 26 (92.9) | 23.8 (3.0–200) | .003 | 0.38 (0.008–18.18) | .625 |
| N = 44 | n = 3 (30.0) | |||||
| Time duration to ocular symptoms (days) | Mean (SD) | 4.4 (±3.0) | 0.8 (0.7–1.0) | .015 | 0.74 (0.47–1.165) | .193 |
| Presence of hypopyon | Y = 41 | n = 38 (92.7) | 19.2 (3.2–111) | .001 | 52.6 (1.7–1000) | .024 |
| N = 12 | n = 5 (41.7) | |||||
| Increased IOP c | Y = 11 | n = 11 (100) | 0 | .999 | ||
| N = 42 | n = 32 (76.2) | |||||
| Presenting VA d | Y = 48 | n = 39 (81.3) | 1.9 (0.03–13.0) | .521 | ||
| N = 6 | n = 4 (66.7) | |||||
| Diabetes mellitus | Y = 34 | n = 28 (82.4) | 1.3 (0.4–4.5) | .662 | ||
| N = 37 | n = 27 (73.0) | |||||
| Intravitreal antibiotics | Y = 56 | n = 44 (78.6) | 1.2 (0.3–4.5) | .818 | ||
| N = 15 | n = 11 (73.3) | |||||
| Vitrectomy performed | Y = 19 | n = 15 (78.9) | 1.1 (0.3–4.1) | .890 | ||
| N = 52 | n = 40 (76.9) | |||||
| Risk factors for evisceration | ||||||
| Patients seen before 2000 | Y = 27 | n = 12 (44.4%) | 4.2 (1.4–12.8) | .013 | 5.2 (1.5–17.9) | .010 |
| N = 44 | n = 7 (16.0%) | |||||
| Unilateral involvement | Y = 51 | n = 18 (35.3%) | 10.4 (1.3–83.3) | .029 | 10.4 (1.2–90.2) | .037 |
| N = 20 | n = 1 (5.0%) | |||||
| Panophthalmic involvement b | Y = 28 | n = 6 (21.4%) | 0 | .999 | ||
| N = 10 | n = 0 (0%) | |||||
| Time duration to ocular symptoms (days) | Mean (SD) | 4.4 (±3.0) | 1.0 (0.9–1.1) | .730 | ||
| Presence of hypopyon | Y = 41 | n = 8 (19.5%) | 2.8 (0.3–25.6) | .367 | ||
| N = 12 | n = 1 (8.3%) | |||||
| Increased IOP c | Y = 11 | n = 2 (18.2%) | 1.1 (0.2–6.3) | .917 | ||
| N = 42 | n = 7 (16.7%) | |||||
| Presenting VA d | Y = 48 | n = 9 (18.8%) | 2.0 (0.3–13.2) | .494 | ||
| N = 6 | n = 0 (0%) | |||||
| Diabetes mellitus | Y = 34 | n = 8 (23.5%) | 0.5 (0.2-1.7) | .294 | ||
| N = 37 | n = 11 (29.7%) | |||||
| Intravitreal antibiotics | Y = 56 | n = 19 (33.9%) | 0 | .998 | ||
| N = 15 | n = 0 (0%) | |||||
| Vitrectomy performed | Y = 19 | n = 9 (47.4%) | 3.7 (1.1–11.9) | .030 | 0.4 (0.1–1.6) | .215 |
| N = 52 | n = 10 (19.2%) | |||||
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
