To explore the incidence and visual outcomes of acute-onset endophthalmitis after transconjunctival microincision vitrectomy surgery (MIVS).
Retrospective, interventional, multicenter survey with a systematic review.
A clinical database search was performed at 27 institutions involving 43 868 consecutive patients who underwent vitrectomy between November 2003 and October 2008 to identify all patients with endophthalmitis after vitrectomy. A systematic review of studies reporting the endophthalmitis rates after MIVS versus 20-gauge vitrectomy was conducted to assess the pooled incidence rates of postvitrectomy endophthalmitis.
The endophthalmitis rates from the multicenter survey were 0.034% (10 cases per 29 030 eyes) after 20-gauge vitrectomy and 0.054% (8 cases per 14 838 eyes) after MIVS, with no significant ( P = .603) differences between groups. Although the incidence in 25-gauge cases (6 per 8238 eyes; 0.073%) was greater than in 23-gauge cases (2 per 6600 eyes; 0.030%), the difference was not significant ( P = 0.451). Of 8 eyes in which endophthalmitis developed after MIVS, 6 eyes (75%) had a final visual acuity of 0.5 or better, and none lost light perception. By combining the results of 7 studies, including the current multicenter survey, meta-analyses from a total of 77 956 cases at the baseline showed that the pooled endophthalmitis rates after MIVS (0.08%; 95% confidence interval, 0.030% to 0.164%) and after 20-gauge vitrectomy (0.030%; 95% confidence interval, 0.012% to 0.048%) did not differ significantly ( P = .207, pooled risk difference; 0.0005 [95% confidence interval, −0.0002 to 0.0012]).
The incidence of postvitrectomy endophthalmitis was low with no significant differences between MIVS and 20-gauge vitrectomy.
Acute-onset postvitrectomy endophthalmitis is uncommon but remains one of the most serious complications associated with devastating visual loss or blindness, despite appropriate treatment. The reported incidence of infectious endophthalmitis after conventional 20-gauge vitrectomy has decreased to 0.03% to 0.05% during the past decades because of improved surgical techniques and recognition of the importance of perioperative antiseptic preparations. However, because of the recent trend toward more frequent use of and expanding indications for microincision vitrectomy surgery, there is a growing concern that microincision vitrectomy surgery with either 23- or 25-gauge instrumentation may increase the risk of postoperative endophthalmitis compared with conventional 20-gauge vitrectomy because of the transconjunctival approach and sutureless nature of the procedure.
Based on a computerized search of PubMed (National Library of Medicine) databases from 2002 when modern microincision vitrectomy surgery began to be performed, acute-onset infectious endophthalmitis after microincision vitrectomy surgery first was reported in 2005 in a case treated with a 25-gauge system, and this was followed by an increasing number of small case series describing the potential risk over the years. To date, several studies have reported the incidence of acute-onset endophthalmitis after transconjunctival 23- or 25-gauge microincision vitrectomy surgery compared with that after conventional 20-gauge vitrectomy. However, the reported incidence of acute-onset endophthalmitis after microincision vitrectomy surgery varies across studies (0% to 1.55%), because most reports are based on the experience of individual institutions or groups of surgeons and because statistically valid data often are limited by small sample sizes. Although a prospective, randomized, controlled study clearly would be most valid for determining the accurate incidence rates of endophthalmitis after microincision vitrectomy surgery compared with conventional 20-gauge vitrectomy, the relative rarity of endophthalmitis poses a significant challenge when conducting such a clinical trial because extraordinarily large numbers of patients (no fewer than 30 000 per group) would be required for robust data in prospective studies, given the low incidence rates. In contrast, as evidenced in the postcataract endophthalmitis literature, other appropriate methods, such as a systematic review of studies or a multicenter survey, are more practical and can identify the clinical and statistical trends of this devastating complication.
In the current study, we conducted a multicenter survey and a systematic overview of the recent literature and combined the results of multiple studies to obtain the best available perspective of the recent trends in acute-onset endophthalmitis after microincision vitrectomy surgery. The clinical features, management strategies, causative organisms, and visual outcomes in patients in whom acute-onset endophthalmitis developed after microincision vitrectomy surgery were investigated.
A Retrospective, Multicenter Survey of Acute-Onset Postvitrectomy Endophthalmitis
A retrospective database search was performed at 27 institutions (Appendix) in Japan to identify all patients who underwent conventional 20-gauge pars plana vitrectomy (PPV) or transconjunctival microincision vitrectomy surgery with 23- or 25-gauge instruments between January 1, 2004, and December 31, 2008, and subsequently were treated for acute-onset infectious endophthalmitis after the initial vitrectomy. The time chosen for a database search was immediately after the introduction of a 25-gauge transconjunctival microincision vitrectomy surgery system in Japan. The search combined the procedure key words of pars plana vitrectomy with the diagnosis key word of endophthalmitis and specified by checking that the date of development of endophthalmitis was within 6 weeks after the date of vitrectomy. The patients’ medical records were reviewed to reconfirm that endophthalmitis was associated with vitrectomy and to collect clinical data to meet the study objectives. The exclusion criteria were a preoperative diagnosis of current intraocular infection or inflammation, such as endogenous endophthalmitis, acute retinal necrosis, cytomegalovirus retinitis, and vitreous opacity associated with uveitis of unknown cause or suspicious malignancy.
In the current multicenter survey, retina specialists performed all surgeries (n = 31) in the 27 institutions after standardized preoperative antiseptic preparation that included topical administration of a third- or fourth-generation fluoroquinolone 3 to 4 times daily beginning 3 days before surgery. After scrubbing the eyelids and periorbital area with 1.25% povidone–iodine solution, undiluted povidone–iodine solution was used to sterilize the surgical field. All patients were draped in the standard fashion with the lashes everted from the surgical field. No antibiotic irrigation or intravitreal antibiotics were used during surgery. At the end of surgery, subconjunctival antibiotics, a steroid, or both were administered at the discretion of the surgeon. A topical antibiotic ointment was applied, and the eye was patched and shielded. Patients routinely received postoperative topical fluoroquinolone and prednisolone acetate for at least 1 month.
In eyes undergoing conventional 20-gauge 3-port vitrectomy, all sclerotomies were created with the 20-gauge microvitreoretinal blade after conjunctival peritomy. At the end of surgery, all 20-gauge sclerotomies and conjunctival incisions were sutured. In contrast, all sclerotomies in eyes undergoing microincision vitrectomy surgery were created transconjunctivally using 23- or 25-gauge trocar–cannula system (Alcon Laboratories, Inc, Fort Worth, Texas, USA; Dutch Ophthalmic Research Center, Zuidland, The Netherlands). The 25-gauge sclerotomies were performed using perpendicular incisions in the early years after the introduction of the 25-gauge system and then with an oblique incision from 2005 onward, whereas the 23-gauge sclerotomies were created by an oblique incision from the time the procedure was introduced. The conjunctiva was displaced before the trocar–cannula was inserted in most cases to misalign the conjunctival and scleral incisions. All cannulas were removed at the end of surgery. The sclerotomies were sutured if the operating surgeon suspected that a leak was present.
Acute-onset postoperative endophthalmitis was defined by the presence of anterior chamber inflammatory cells with fibrin formation, hypopyon, vitreous cells and fibrin, progressive ocular pain, conjunctival congestion, and blurred vision occurring within 6 weeks after surgery. Specimens from the vitreous biopsy were cultured, but the results were unnecessary for diagnosis of acute postoperative endophthalmitis, given the reported high false-negative rate of intravitreal sampling. Data collected from the patients with acute-onset endophthalmitis included demographic information, presenting signs and symptoms, duration of symptoms, presenting visual acuity (VA), surgical details, therapeutic interventions, and postoperative course, including intraocular pressure on the first postoperative day, culture results, and VA outcomes.
Study Search and Selection for Meta-analysis
All studies published in the English literature on endophthalmitis after 23- or 25-gauge microincision vitrectomy surgery were retrieved systematically from MEDLINE, Embase, and the Cochrane Library through November 2009 using the search terms endophthalmitis, incidences or rates, vitrectomy, 23-gauge, and 25-gauge. A manual search also was performed by checking the reference lists of all retrieved trials to identify studies not included in the computerized databases. Only the clinical studies that directly compared the postvitrectomy endophthalmitis rates between microincision vitrectomy surgery (23- or 25-gauge) and conventional 20-gauge vitrectomy were included. To avoid acknowledged and converted data duplication from the same group of patients, only the most recent series or largest study group was included for subsequent analysis.
For the data derived from the current survey, the incidence of acute-onset endophthalmitis associated with microincision vitrectomy surgery and conventional 20-gauge vitrectomy were compared using the chi-square test, and confidence intervals (CIs) were calculated by the percentage of binomial distribution. Other proportions also were compared using the Fisher exact test. The VA was measured using the Landolt C acuity chart and was analyzed on a logarithm of the minimal angle of resolution (logMAR) scale. Counting fingers vision was defined as 0.01 (2.0 logMAR), hand movements vision was defined as 0.005 (2.3 logMAR), light perception vision was defined as only 0.002 (2.7 logMAR), and no light perception vision was defined as 0.001 (3.0 logMAR). Visual improvement was defined as an increase of at least 0.3 logMAR unit. Where appropriate, the Student t test and the Wilcoxon rank-sum test were used to compare the differences between groups.
After study selection and data extraction, dichotomous data from each study eligible for meta-analysis were summarized as risk differences. Considering the heterogeneity among studies, we analyzed the homogeneity statistics and then combined data using a random effects model by restricted the maximum likelihood method to achieve more conservative estimates. The risk differences are presented with 95% CIs.
All primary analyses were performed using either JMP software version 8.0 for windows (SAS Institute, Inc, Cary, North Carolina, USA) or the meta-analyses were carried out with a commercially available software programmed by Kenichi Masui (Shinko Trading Co Ltd, Publication Department, Tokyo, Japan) with Excel. P < .05 was considered statistically significant.
Incidence of Postvitrectomy Endophthalmitis in the Current Survey
A total of 43 868 vitrectomies met the inclusion criteria during the study period: 29 030 cases treated with conventional 20-gauge vitrectomy, 8238 cases treated with the 25-gauge system, and 6600 cases treated with the 23-gauge system. Of these, acute-onset postvitrectomy endophthalmitis developed in 18 eyes (0.041%; 95% CI, 0.022% to 0.060%); the incidence of acute infectious endophthalmitis was 0.034% (10 cases per 29 030 eyes; 95% CI, 0.013% to 0.056%) for conventional 20-gauge PPV compared with 0.054% (8 cases per 14 838 eyes; 95% CI, 0.017% to 0.091%) for microincision vitrectomy surgery. The difference between the procedures was not significant ( P = .603). Of the 14 838 eyes treated with microincision vitrectomy surgery, the incidence of endophthalmitis among the 25-gauge cases (6 per 8238 eyes; 0.073%) was more than twice as high as that among the 23-gauge cases (2 per 6600 eyes; 0.030%), but the difference did not reach significance ( P = .451). The mean follow-up period was 25.3 ± 11.8 months (range, 6 to 60 months).
Systematic Overview of Differences in Endophthalmitis Frequency between Microincision Vitrectomy Surgery and Conventional 20-Gauge Vitrectomy
Seven potentially relevant publications were identified in the literature search. One study was excluded because of possible duplication of patient data from another large series. After adding the data from the current multicenter survey, 7 retrospective studies met the criteria for subsequent meta-analysis. Because there was heterogeneity among the studies, a random-effects model (restricted maximum likelihood method) was applied for meta-analysis ( Table 1 ). The 7 studies pooled a total of 77 956 eyes, and the pooled estimates of endophthalmitis after microincision vitrectomy surgery (0.080%; 95% CI, 0.030% to 0.164%) were numerically higher than that after conventional 20-gauge vitrectomy (0.030%; 95% CI, 0.012% to 0.048%); however, the difference between the groups was not significant ( P = .207), with a pooled risk difference of 0.0005 (95% CI, −0.0002 to 0.0012).
|Study||Centers||Study Period (yrs)||Transconjunctival MIVS||20-Gauge Vitrectomy||Risk Difference (95% CI)|
|Kunimoto and Kaiser, 2007||1||2004 through 2006||7/3103 a||0.226||1/5498||0.018||0.0021 (0.0004 to 0.0038)|
|Scott and associates, 2008||7||2005 through 2006||11/1307 a||0.842||2/6375||0.031||0.0081 (0.0031 to 0.0131)|
|Simada and associates, 2008||1||2000 through 2007||1/3343 a||0.030||1/3592||0.028||0.0000 (−0.0008 to 0.0008)|
|Chen and associates, 2009||1||2002 through 2006||1/431 a||0.232||1/3046||0.033||0.0020 (−0.0026 to 0.0066)|
|Hu and associates, 2009||1||2002 through 2006||1/1424 a||0.070||0/1948||0||0.0007 (−0.0007 to 0.0021)|
|Parolini and associates, 2009||1||2003 through 2008||0/943 b||0||1/3078||0.032||−0.0003 (−0.0010 to 0.0003)|
|Current study||27||2004 through 2008||8/14 838 c||0.054||10/29 030||0.034||0.0002 (−0.0002 to 0.0006)|
|Pooled estimates||0.080 d||0.030 d||0.0005 (−0.0002 to 0.0012)|
|Test for homogeneity||chi-square = 18.21, df = 6, P = .0057|
|Test for difference in incidence||Z = 1.59, P = .207|
c Two (0.030%) of 6660 eyes with the 23-gauge system; 6 (0.073%) of 8238 eyes with the 25-gauge system.
d Weighted incidence of postvitrectomy endophthalmitis was derived from the pooled data of 7 studies.
Treatment Outcomes of Postvitrectomy Endophthalmitis in the Current Survey
The patient baseline characteristics and surgical details of the 18 eyes with postvitrectomy endophthalmitis in the current multicenter survey are shown in Table 2 . All eyes were treated with topical antibiotics at the time of presentation. Of the 8 patients (5 women, 3 men) in whom acute-onset endophthalmitis developed after microincision vitrectomy surgery, 2 patients (25%) had diabetes mellitus. All surgeries were uneventful: 4 eyes with macular diseases underwent simple core vitrectomy, and 4 eyes underwent extensive vitreous removal with peripheral vitreous shaving. Seven eyes (87.5%) were left with a fluid-filled vitreous cavity at the end of surgery, whereas fluid-air exchange followed by long-acting gas tamponade was performed in one eye (12.5%). All sclerotomies except one were self-sealing. One eye (12.5%) had postoperative hypotony (intraocular pressure, ≤ 7 mm Hg) from day 1 after surgery. Of the 10 patients (4 women, 6 men) in whom acute-onset endophthalmitis developed after conventional 20-gauge PPV, 6 patients (55%) had diabetes mellitus. All surgeries were uncomplicated; 80% of patients underwent extensive vitreous removal with peripheral vitreous shaving. Eight eyes (80%) were left with a fluid-filled vitreous cavity at the end of surgery, whereas silicone oil tamponade was performed in 2 eyes (20%).
|Case No.||Age (yrs)||Gender||Eye||Preoperative Diagnosis||Preoperative VA||Lens Status||Gauge for Surgery||Vitrectomy a||No. of Sutured Sclerotomies||Vitreous Tamponade||Surgical Time (min)||Postoperative Hypotony b||Systemic Medical Factors|
|1||69||M||R||VH, PDR||0.05||Phakic||20||Extensive||3||Fluid||72||No||Diabetes mellitus, renal dialysis|
|4||53||M||L||VH, PDR||0.4||Phakic||20||Extensive||3||Silicone oil||60||No||Diabetes mellitus|
|5||63||M||R||VH, PDR||0.1||IOL||20||Extensive||3||Fluid||90||No||Diabetes mellitus|
|6||59||F||R||TRD, PDR||0.5||Phakic||20||Extensive||3||Fluid||110||No||Diabetes mellitus|
|7||63||F||R||VH, PDR||0.1||Phakic||20||Extensive||3||Fluid||105||No||Diabetes mellitus, hypertension|
|8||41||M||L||TRD, PDR||0.06||Phakic||20||Extensive||3||Fluid||150||No||Diabetes mellitus|
|10||57||M||L||Dislocated lens fragment||0.03||Aphakic||20||Core||3||Fluid||110||No|
|18||65||F||L||VH, PDR||HM||IOL||23||Extensive||0||Fluid||40||No||Diabetes mellitus|
a Vitrectomy with peripheral vitreous shaving was defined as extensive; vitrectomy limited to the central portion without peripheral vitreous shaving was defined as a core vitrectomy.
b Postoperative hypotony was defined as intraocular pressure ≤ 7 mm Hg observed within 1 week after surgery.
The presenting characteristics and treatment results of the 18 eyes are shown in Table 3 . Of the 8 patients who underwent microincision vitrectomy surgery, the median time between surgery and presentation with endophthalmitis was 2 days (mean, 1.9 days; range, 1 to 3 days), and 63% of patients had blurred vision rather than ocular pain as the initial manifestation. Seven (87.5%) of the 8 eyes underwent vitrectomy with intraocular irrigation with antibiotics; the other eye underwent a tap and injection procedure with intravitreal antibiotics because of relatively mild intraocular inflammation. Vancomycin was used intravitreally in all 8 eyes, and ceftazidime was used in 7 eyes (87.5%). Culture-positive organisms were detected in 5 eyes (63%), whereas the other 3 cases were culture negative. All positive-culture organisms responded to intraocular antibiotics. The presenting VA was available for 5 eyes: hand movements (n = 2), counting fingers (n = 1), 0.1 (n = 1), and 0.03 (n = 1). After treatment, 6 eyes (75%) had a VA of 0.5 (20/40) or better and 5 eyes (63%) had visual improvement compared with that before surgery. Of the 10 patients with endophthalmitis after 20-gauge PPV, the median time between surgery and presentation with endophthalmitis was 2 days (mean, 4.6 days; range, 1 to 23 days). Eight (80%) of the 10 eyes underwent revisions with intraocular irrigation with antibiotics, and 2 eyes underwent a tap and injection procedure with intravitreal antibiotics because of relatively mild intraocular inflammation. Vancomycin was used intravitreally in 8 eyes, ceftazidime was used in 5 eyes, and imipenem was used in 2 eyes. Organisms were identified in 3 (37.5%) of the 8 eyes, and the other 5 cases were culture negative. All culture-positive organisms responded to the intraocular antibiotics. The presenting VA levels were available for 7 eyes: light perception (n = 2), hand movements (n = 3), 0.1 (n = 1), and 0.03 (n = 1). After treatment, 2 eyes (20%) had a VA of 0.5 or better, and 1 eye lost light perception. The VA improved in 3 eyes (30%) compared with that before surgery.
|Case No.||Days to Presentation||Presenting VA||Main Symptoms at Presentation||Clinical Findings||Organism||Intraocular Treatment||Final VA|
|1||5||NA||Blurred vision||Fibrin, vitreitis||Culture negative||Vitrectomy, irrigation with vancomycin, imipenem||0.1|
|2||1||NA||Ocular pain||Fibrin, hypopyon, retinal hemorrhages||Culture specimen not obtained||Tap with vancomycin||1.0|
|3||2||HM||Blurred vision||Cells, vitreitis, retinal necrosis||Culture negative||Vitrectomy, irrigation with imipenem||0.7|
|4||2||HM||Blurred vision||Fibrin, hypopyon||Culture negative||Vitrectomy, irrigation with imipenem||0.2|
|5||4||0.03||Blurred vision||Cells, fibrin, hypopyon, vitreitis, retinal vasculitis||MRSE||Vitrectomy, irrigation with vancomycin, ceftazidime, IOL extraction||0.3|
|6||2||NA||Ocular pain||Cells, fibrin, hypopyon, vitreitis||Culture negative||Vitrectomy, irrigation with vancomycin, ceftazidime||0.1|
|7||2||LP||Ocular pain||Cells, fibrin, hypopyon, vitreitis||Pneumococcus||Vitrectomy, irrigation with vancomycin, ceftazidime||HM|
|8||2||LP||Ocular pain||Cells, fibrin, hypopyon, vitreitis, retinal necrosis||MRSA||Vitrectomy, irrigation with vancomycin, ceftazidime||NLP|
|9||3||0.1||Blurred vision||Cells, fibrin, vitreitis||Culture specimen not obtained||Tap with vancomycin||0.8|
|10||23||HM||Blurred vision||Cells, fibrin, vitreitis||Culture negative||Vitrectomy, irrigation with vancomycin, ceftazidime||LP|
|11||1||HM||Ocular pain||Cells, fibrin, hypopyon, retinal hemorrhages||MRSE||Vitrectomy, irrigation with vancomycin, ceftazidime||0.1|
|12||3||CF||Blurred vision||Cells, fibrin, hypopyon, retinal hemorrhages||MRSE||Vitrectomy, irrigation with vancomycin, ceftazidime||1.0|
|13||2||NA||Blurred vision||Cells, fibrin, hypopyon, retinal hemorrhage with vasculitis||Propionibacterium acnes||Vitrectomy, irrigation with vancomycin, ceftazidime||0.9|
|14||2||0.03||Blurred vision||Cells, fibrin, vitreitis||Culture negative||Vitrectomy, irrigation with vancomycin, ceftazidime||0.7|
|15||2||0.1||Blurred vision||Cells, fibrin, vitreitis||Culture negative||Vitrectomy, irrigation with vancomycin, ceftazidime||0.6|
|16||2||NA||Blurred vision||Cells, fibrin, vitreitis||Culture negative||Vitrectomy, irrigation with vancomycin, ceftazidime||1.0|
|17||2||NA||None||Cells, fibrin, hypopyon, retinal hemorrhages||Enterococcus faecalis||Vitrectomy, irrigation with vancomycin, imipenem||0.7|
|18||1||HM||None||Cells, fibrin, hypopyon||MSSA||Tap with vancomycin, ceftazidime||0.1|