Article Text

Download PDFPDF

Occurrence and risk factors for retinal detachment after pars plana vitrectomy in acute postcataract bacterial endophthalmitis
  1. Christophe Chiquet1,2,
  2. Florent Aptel1,2,
  3. Aurélie Combey-de Lambert1,2,
  4. Alain M Bron3,
  5. Nelly Campolmi4,
  6. Karine Palombi1,2,
  7. Gilles Thuret4,
  8. Frédéric Rouberol5,
  9. Pierre-Loïc Cornut5,
  10. Catherine Creuzot-Garcher3
  11. On behalf of the FRIENDS (French Institutional Endophthalmitis Study) group
    1. 1Grenoble Alpes University, Grenoble, France
    2. 2Department of Ophthalmology, University Hospital, Grenoble, France
    3. 3Department of Ophthalmology, University Hospital, Dijon, France
    4. 4Department of Ophthalmology, University Hospital, Saint-Etienne, France
    5. 5Department of Ophthalmology, Edouard Herriot University Hospital, University of Lyon I, Lyon, France
    1. Correspondence to Professor Christophe Chiquet, Department of Ophthalmology, University Hospital, BP 217, Grenoble 38043, France; cchiquet{at}


    Background/aims To report the incidence, risk factors and prognosis of retinal detachment (RD) in patients who had vitrectomy for acute bacterial endophthalmitis after cataract surgery.

    Methods 123 patients with acute postcataract endophthalmitis, consecutively treated with pars plana vitrectomy (PPV) were included by the French Institutional Endophthalmitis Study group, in a prospective multicentre cohort study. Risk factors of RD were analysed using logistic regression.

    Results At the 6-month follow-up, the rate of post-PPV RD was 13% (n=16). The risk factors of post-PPV RD were diabetes (OR=4.7 (1.4–15.4), p=0.01) and visualisation of retinal vasculitis on the posterior pole (OR=3.8 (1.1–13.9), p=0.03) at the time of PPV. Postoperative RD occurred in 56% (n=9) of cases in the first month, in 31% (n=5) in the second month and in 6% (n=1) in the third month, with a mean delay of 47±71 days after PPV. The macula was detached in 12 cases (75%) and proliferative vitreoretinopathy grade C was present in seven cases. Final successful reattachment of the retina was obtained in 60% (n=9/15) of cases, with one (7/9) or two surgeries (2/9). Final visual acuity after surgical repair was ≥20/40 in 19% of cases, compared with 43% in patients without RD (p=0.05).

    Conclusions RD is a major and severe complication of PPV performed in patients with acute postcataract endophthalmitis. Retinal vasculitis is a major risk factor of RD after PPV. Anatomical and functional outcome remain poor.

    • Infection
    • Inflammation
    • Treatment Surgery
    • Retina
    • Microbiology

    Statistics from


    Postcataract endophthalmitis is a rare (approximately 0.08% of cataract surgeries)1 ,2 but serious complication, which can lead to anatomical and/or functional loss of the eye. Pars plana vitrectomy (PPV) is used for vitreous sampling for microbiological analysis, removal of intracavital accumulation of purulent material with bacteria and toxins, improvement of the distribution of antibiotics in the vitreous cavity and cleaning of vitreous opacities and membranes, which can induce pathological vitreoretinal adhesion.3 On the other hand vitrectomy may reduce the half-life of vancomycin in the eye4 and may prove to be very difficult, especially in case of poor visibility caused by corneal opacity, posterior synechia, capsular bag infiltration and pupillary fibrin membranes.

    The rate of retinal detachment (RD) after PPV performed for endophthalmitis remains high (RD, 5%–21%).5–8 Few data are available concerning the risk factors and final prognosis of patients with endophthalmitis complicated by RD.5 ,6 ,8 In the only prospective study conducted to date, the Endophthalmitis Vitrectomy Study (EVS),8 RD was associated with virulent bacteria, capsular rupture and an early additional procedure. At the time of the EVS, these RDs carried a poor visual prognosis, with 27% of patients achieving 20/40 final visual acuity (VA).8 Further data are necessary because of the restrictive exclusion criteria of the EVS (patients willing to be randomised, exclusion of patients with initial VA equal to or better than 20/100 and good media clarity) and since PPV has benefitted from many technical advances, especially in fundus-viewing devices and higher cutting rates.

    The main objectives of this prospective multicentre study were to evaluate the incidence and the risk factors of RD after vitrectomy in patients with acute postcataract endophthalmitis and to analyse final prognosis of these patients with RD.

    Patients and methods

    The French Institutional Endophthalmitis Study group included 233 patients with acute postcataract endophthalmitis between March 2004 and December 2011 in four French university hospitals. In this population, 91 patients had one or more intravitreal injections of antibiotics. Three per cent of these 91 patients (n=3) developed an RD in the first 6 months of follow-up. One hundred and thirty-four patients (134 eyes, 57%) were treated with PPV. Eleven out of the 134 patients who had vitrectomy were lost to follow-up (within 6 months). Finally, 123 eyes of 123 patients with baseline and 6-month evaluation were analysed in this report.

    The study followed the Declaration of Helsinki guidelines for research involving human subjects and was approved by the local institutional review board (IRB# 5921).

    Patients were eligible if they had displayed clinical signs of endophthalmitis within 6 weeks after cataract surgery9 and would benefit from PPV. The inclusion criteria, data sheet, sampling techniques, microbiological procedures, antibiotics used for intravitreal injections and follow-up visits were standardised in the four centres.

    Fundus visibility was defined upon visualisation of second-order retinal vessels.9 A good visual outcome was defined as a final VA ≥20/40.

    PPV was indicated when VA was limited to light perception or hand motion, or in case of rapid anatomical (increase in corneal oedema, hypopyon and/or hyalitis within 48 h) or functional deterioration (VA going to light perception or lower). PPV was, therefore, performed at admission or after one intravitreal injection of vancomycin (1 mg/0.1 mL) and ceftazidime (2.225 mg/0.1 mL). The need for PPV was assessed on admission and daily. The Accurus vitrectomy system (Alcon Laboratories, Rueil-Malmaison, France) was used in all centres with a cutting rate of 1500–2500/min and a suction rate of 120–400 mL/min. A contact wide-angle viewing system combined with an image inverter (SDI Oculus IIe, Wetzlar, Germany) was used for fundus visualisation.

    Microbiological data

    Ocular samples were obtained from an immediate tap of aqueous humour (AH) and/or vitreous at the time of intravitreal injections of antibiotics, or from vitreous sampling at the beginning of PPV, as previously reported.10 One-half of the vitreous or AH specimens was directly injected into a brain heart infusion broth (BHI, 10 mL, pH 7.4±0.2, ref. ADM 88440, AES Laboratories, Combourg, France) for conventional culture. The other half was stored in a DNA-free microcentrifuge tube for PCR. Culture and PCR are reported in details in previous reports.10–14 Patients were then further classified into the following groups according to microbiological results: detection of no microorganisms, coagulase negative Staphylococci species (CNSPs) or more virulent species (including Streptococcus species, Staphylococcus aureus, S. lugdunensis and gram-negative bacteria).13

    Statistical analysis

    Statistical analysis was performed using the Statistical Package for the Social Sciences program (SPSS V.17.0 for Windows, Chicago, Illinois, USA). Quantitative data were expressed as mean±SD. A logistic regression model was used to estimate ORs with 95% CIs of quantitative factors or qualitative factors. The tests were two-tailed and statistical significance was set at p<0.05.


    One hundred and twenty-three patients with acute postcataract endophthalmitis and treated with PPV within 10 days after admission were analysed. There were 62 men and 61 women with a mean age of 74±11 years. Complicated cataract surgery (especially capsular rupture) was reported in 18% of cases. The mean delay between the onset of symptoms and PPV was 3.3±2 days. Microbiological identification is summarised in table 1.

    Table 1

    Bacterial identification in patients with acute endophthalmitis

    Pars plana vitrectomy

    An immediate antibiotic intravitreal injection was the first-line treatment for 90% of the patients before PPV, which was performed after one (50%), two (35%) or three (5%) intravitreal injections. After presentation, the median delay until PPV was 2 days (range, 0–10 days). At admission, VA limited to light perception was noted in 49% of the cases. Rapid anatomical deterioration and functional deterioration were noted in 44 patients (36%) and 39 patients (32%), respectively.

    General anaesthesia was used in 90% of the cases. Most patients underwent a standardised vitrectomy (23-G transconjunctival vitrectomy in seven cases). The mean duration of PPV was 60±23 min. Capsulectomy was performed in 20 cases (16%). The intraocular lens was removed in one case due to a subluxation in the vitreous cavity. PPV was incomplete in 64% of the cases and posterior vitreous detachment was actively performed in seven patients. Retained lens fragments in the vitreous were removed in one case. At the end of PPV, intravitreal injections of antibiotics (using a 1:5 dilution, ie, vancomycin (0.2 mg/0.1 mL) and ceftazidime (0.4 mg/0.1 mL)) and corticosteroids were given in 96% and 65% of cases, respectively.

    During PPV, the retina was: (1) visible and apparently normal (25%, n=29), (2) visible and abnormal (52%, n=63) or (3) not visible (defined by insufficient visualisation of the retina and/or visualisation of <50% of the retina and/or the posterior pole, 23%, n=31). Signs of vasculitis (retinal vascular sheathing, retinal haemorrhages) were noted on the posterior pole in 34 patients and on the mid-periphery in 30 patients. Fundus examination confirmed the presence of one case of pre-existing RD and five choroidal detachments, previously diagnosed on ultrasound examination done before surgery. During PPV, complications occurred in 10% of the 123 eyes, with at times two complications per eye: retinal tears (n=3, one of which developed an RD), intraocular lens luxation (n=1), choroidal detachment (n=4), choroidal haemorrhage (n=1) and intravitreal haemorrhage (n=4). One case of suprachoroidal infusion was noted at the infusion line site. One case of posterior retinal tear occurred during PPV without accompanying posterior vitreous detachment in a patient with severe retinitis.

    Functional and anatomical prognosis at 6 months

    At the 6-month follow-up (table 2), 49 patients (40%) had a VA ≥20/40 and 83 patients (68%) ≥20/200. Ocular complications were mainly phthisis (n=10, 8%), RD (n=16, 13%), epiretinal membrane (n=19, 16%), macular oedema (n=14, 12%) and/or choroidal detachment (n=6, 5%).

    Table 2

    Anatomical and visual outcome in 123 patients who had vitrectomy following acute postcataract endophthalmitis

    Note: some data are missing for the following criteria: corneal oedema (n=1), posterior synechia (n=5), retinal tears (n=2), epiretinal membrane (n=4), macular oedema (n=6) and choroidal detachment (n=1). For these criteria, the total number of patients noted is, therefore, <123.

    RD after PPV

    Almost all RDs were reported after PPV (16/17 cases) and one case of RD was diagnosed by ultrasound examination before PPV. This RD was total and caused by an inferior giant retinal tear. This case was treated using PPV, perfluorocarbon liquid, endolaser and heavy silicone oil. The final prognosis was poor with light perception at 6 months and the persistence of subretinal fluid in the inferior retina.

    Table 3

    Characteristics of retinal detachment occurring in patients who had vitrectomy following acute postcataract endophthalmitis

    Risk factors of RD (based on the occurrence of 16 RDs after PPV) occurring during or after PPV were diabetes (OR=4.7 (1.4–15.4), p=0.01) and visualisation of retinal vasculitis on the posterior pole (OR=3.8 (1.1–13.9), p=0.03) at the time of PPV. One patient who underwent intraocular lens removal during PPV developed an RD. Intraoperative complications during cataract surgery, age, sex, axial length >26 mm, baseline VA, inflammation status at baseline (hypopyon, pupillary fibrin membrane, fundus visibility, corneal oedema), microbiological factors (virulence or bacterial identification), type of PPV (complete or incomplete) and intravitreal injection of steroids at the end of PPV were not significantly associated with RD occurrence. Surgical detachment of the posterior hyaloid during PPV was rare (7/123).

    The characteristics of RD are summarised in table 3. Postoperative RD occurred with a mean delay of 47±71 days after PPV. RD was due to one or more superior tears in 50% of cases; these were not related to the sclerotomy sites. In 10 out of 16 cases, tears were identified before surgery. During surgery (n=15), retinal tears were not identified in three cases; one retinal tear was noted in seven patients, two retinal tears in three cases and three retinal tears in one patient. Two giant tears were reported.

    One patient refused RD surgery (total RD with absence of light perception, no bacteria identified) and a few months later underwent enucleation. Of the 15 operated patients, 11 had one surgery and four patients underwent two RD surgeries. For the first RD surgery, PPV was performed in 15 cases, associated with cryotherapy (n=8) or endolaser (n=7) and scleral buckling (n=5, 360° in two out of five cases). Intraocular tamponade was done with silicone oil (n=9) or gas (n=6). Epiretinal membranes were peeled in four cases. Two patients needed large retinectomy (180°) and two others underwent 360° laser photocoagulation. After one surgery, seven patients had reattached retina and eight patients had unsuccessful RD surgery (figure 1). Of these eight patients, four patients agreed to a second RD surgery (using PPV, endolaser and silicone tamponade), which was successful in two out of the four cases. Final successful reattachment of the retina was obtained in 60% (n=9/15) of cases.

    Figure 1

    Follow-up of patients with RD after pars plana vitrectomy for acute postcataract endophthalmitis. CF, count fingers; HM, hand motion; RD, retinal detachment.

    In patients with detached retina at 6 months, RD surgery was performed once (7/9 cases) or twice (2/9). Final VA after surgical repair was ≥20/40 in 19% of cases, compared with 43% in patients without RD (p=0.05). Conversely, final VA was ≤20/100 in 75% of patients operated on for RD, versus 34% of patients without RD (p=0.002).

    Four severe cases of hypotony (intraocular pressure (IOP) <5 mm Hg) and one case of intraocular hypertension IOP >21 mm Hg) were noted. Two additional cases of phthisis occurred: one after RD surgery, the other in a patient who refused RD surgery.


    This prospective study showed that diabetes and retinal vasculitis were the main risk factors of RD in patients who had vitrectomy after acute postcataract endophthalmitis. After RD surgery, the anatomical and functional prognosis remains poor.

    This study provides clinicians with information on the clinical outcome and complications of PPV in a real-life environment. As previously reported in other studies,5 ,15 PPV was performed mostly after at least one intravitreal injection of antibiotics, for cases with poor VA at admission (from light perception to count fingers) and/or rapid degradation of visual or anatomical conditions within the first 24 h. The selection criteria did not differ greatly from those of the EVS,9 except for the exclusion of patients with initial VA equal to or better than 20/100 (n=8 in the current series), the type of surgery (phacoemulsification only) and the delay before PPV immediately after admission. We found no major difference in the proportion of CNSPs between the two studies.

    Surgical techniques should also be discussed when complications of PPV are analysed. The 20-G vitrectomy was the most frequently used technique in the present study, and PPV was considered incomplete in 64% of cases. All retinal tears were identified outside the sclerotomy site. The potential benefit of transconjunctival vitrectomy could be the absence of conjunctival dissection on an inflamed eye, the highest cutter speed of the vitrector, with reduced traction on the retina.16

    In this series, RD (0.8%) was rarely observed at admission, consistent with the results of population screening by the EVS (1.2%).17 RD occurred intraoperatively or postoperatively in 16 out of 17 patients (13% of the overall population), as previously reported (8%–21%).6 ,7 ,18 The risk of RD remains high despite new PPV techniques, especially a higher cutting rate. Undiluted vitreous sampling at the start of PPV for endophthalmitis leads to hypotony, with a potential risk of vitreoretinal tractions and choroidal detachment or haemorrhage. This could be avoided when using diluted vitreous10 since diluted and undiluted vitreous had the same microbiological efficiency using PCR. This surgery may also be complicated intraoperatively by RD.8 ,19 ,20

    In the present study, diabetes and posterior vasculitis were the two main risk factors of RD. The causal relationship between diabetes (without advanced retinopathy) and RD has not been reported before and may be related to the status of the vitreous.21 Posterior vasculitis may be an inflammation marker, and therefore, may reinforce vitreoretinal tractions. Although retinal periphlebitis and retinal haemorrhages have been noted as early signs of bacterial endophthalmitis,22 ,23 the presence of retinal haemorrhages at the time of PPV could be associated with the development of phthisis.24 The risk factors reported in the EVS,9 such as presenting VA, bacterial virulence and open posterior capsule, were not significant in our series. The time to RD occurrence has been reported in the first month (56% in our series, 31% in the EVS),8 the second month (87% vs 43%) and the third month (93% vs 60%). This series differed from the EVS in the rate of RD surgery (93% in our series vs 44% in the EVS), PPV rate (100% vs 57%), reattachment rate (60% vs 78%) and visual outcome (VA >20/40 in 40% vs 33%).

    In the present study, 40% of the patients had a final VA ≥20/40 and 32% <20/400. This is comparable with Lalwani et al,5 who found, in the PPV and inject group, 37% of patients with VA ≥20/40 and 21% with VA <5/200. As shown in our series, intraoperative or postoperative RD remains a significant cause of poor final VA. In our series, the incidence of phthisis bulbi is high (8%) after PPV, compared with 2% in the EVS.

    We acknowledge several limitations to our study inherent to the real-life conditions investigated. First, the timing of PPV could vary for several reasons, poor media clarity, difficult access to the operating room in emergency situations and retinal surgeon availability. Second, the aim of this study was not to define the place of PPV as EVS did, but we focused on severe complications of PPV in patients with acute endophthalmitis. Finally, the EVS criteria were stricter than in the present study in terms of VA, media clarity and the time to PPV, which could explain the results.

    In conclusion, this non-comparative prospective study emphasises the prognostic role of retinal vasculitis and the current risk of RD after PPV in up to 13% of cases. RD surgery can be performed in almost all cases but final anatomical and visual results remain poor, in spite of updated vitreoretinal techniques.


    The authors wish to thank the French Health Ministry (PHRC, French public programme of clinical research), AFAU (French association of unilateral amblyopia), ARFO (association for research in ophthalmology, Grenoble) and Théa for their support. We would also like to acknowledge the work of Zineb Baidi (Clinical Research Associate) and Thierry Zhou (MD) for their contribution to database management.



    • Collaborators THE FRIENDS GROUP, FRench Institutional ENDophthalmitis Study group. Participants: study coordinator: CC. Statistics, methodology, microbiological techniques: François Vandenesch, Max Maurin. Database management: P-LC. Ophthalmology: University Hospital of Dijon: Pierre-Olivier Lafontaine, CC-G, AMB. University Hospital of Grenoble: FA, CC, AC-dL, KP, Thierry Zhou, Eva Brillat-Zaratzian, Jean-Paul Romanet. University Hospital of Lyon (E. Herriot Hospital): P-LC, FR, Philippe Denis. University Hospital of Saint-Etienne: NC, GT, Philippe Gain. Microbiology: University Hospital of Dijon: André Péchinot, Catherine Neuwirth. University Hospital of Grenoble: Sandrine Boisset, Max Maurin. University Hospital of Lyon: Jérôme Etienne, Yvonne Benito Engineer, Anne Tristan, François Vandenesch. University Hospital of Saint-Etienne: Anne Carricajo, Gérard Aubert. Mycology: University Hospital of Dijon: Frédéric Dalle, Alain Bonin. University Hospital of Grenoble: Bernadette Lebeau, Hervé Pelloux. University Hospital of Lyon: Frédérique de Montbrison, Stéphane Picot. University Hospital of Saint-Etienne: Hélène Raberin, Roger Tran Manh Sung.

    • Contributors CC, P-LC, GT and FA designed data collection tools, wrote the statistical analysis plan and analysed the data. CC is guarantor. CC, AC-dL, AMB, NC, KP, GT, FR, P-LC and CC-G included patients and monitored data collection. CC, FA, AMB, NC, GT and CC-G drafted and revised the paper. All authors revised the paper.

    • Funding Health Ministry (PHRC, French public programme of clinical research), AFAU (French association of unilateral amblyopia), ARFO (association for research and learning in ophthalmology) and the Laboratoires Théa.

    • Competing interests None declared.

    • Patient consent Obtained.

    • Ethics approval IRB CPP sud est V (IRB#5921).

    • Provenance and peer review Not commissioned; externally peer reviewed.

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Linked Articles

    • At a glance
      Keith Barton James Chodosh Jost Jonas