Purpose
To determine the incidence, risk factors, and outcomes of delayed suprachoroidal hemorrhage after vitrectomy.
Design
Retrospective multicenter cohort study.
Methods
All consecutive patients who underwent primary vitrectomy, from January 2009 to December 2014, at 4 tertiary vitreoretinal centers in Italy were enrolled. Patient demographics and systemic, ophthalmic, operative, and postoperative data from all centers were extracted from the electronic record system using standardized data collection forms. All eyes that developed delayed suprachoroidal hemorrhage within 48 hours of the end of the vitrectomy were identified as the delayed suprachoroidal hemorrhage group; all other eyes that underwent vitrectomy in the same period, without delayed suprachoroidal hemorrhage, were considered the control group.
Results
From a total of 4852 vitrectomy procedures, 39 cases of delayed suprachoroidal hemorrhage (0.8%) were identified. Multivariable logistic regression showed that significant risk factors for developing delayed suprachoroidal hemorrhage included advancing age (odds ratio [OR], 2.22; P < .001), longer axial length (OR, 2.57; P < .001), presence of rhegmatogenous retinal detachment (OR, 3.27; P = .005), extensive intraoperative photocoagulation (OR, 4.94; P < .001), and emesis postoperatively (OR, 24.39; P < .001). Decision-tree analysis showed that the stronger predictors of delayed suprachoroidal hemorrhage were emesis postoperatively ( P < .001) and extensive intraoperative photocoagulation ( P < .001). After a mean follow-up of 27 ± 8 months, the best-corrected visual acuity decreased from 1.3 preoperatively to 1.6 logarithm of minimal angle of resolution at last follow-up ( P < .001).
Conclusions
Delayed suprachoroidal hemorrhage occurs in 0.8% of vitrectomized eyes. The main risk factors are postoperative emesis and intraoperative extensive photocoagulation.
Suprachoroidal hemorrhage is an uncommon but dramatic condition defined as the presence of blood in the suprachoroidal space as a consequence of rupture of the posterior ciliary arteries or vortex veins. It is considered one of the most potentially devastating complications of all types of intraocular surgery, including vitreoretinal surgery.
When suprachoroidal hemorrhage occurs during surgery it is called “acute intraoperative suprachoroidal hemorrhage,” whereas if it develops during the postoperative period it is called “delayed suprachoroidal hemorrhage.” Acute intraoperative suprachoroidal hemorrhage differs from delayed suprachoroidal hemorrhage in incidence, pathophysiology, and management. Intraoperative suprachoroidal hemorrhage is associated with loss of red reflex and iris and vitreous prolapse, and it can cause extrusion of ocular contents. Suprachoroidal hemorrhage may occur during any intraocular procedure, especially when accompanied by large intraocular pressure fluctuations. Delayed suprachoroidal hemorrhage may occur hours or days after intraocular surgery, and is characterized by sudden severe pain, decreased vision, and a shallow anterior chamber, usually following a Valsalva maneuver–like activity. A recent study showed that the rate of suprachoroidal hemorrhage during vitrectomy was approximately 1%. The risk factors for developing intraoperative suprachoroidal hemorrhage were male sex, advancing age, the use of antiplatelet or anticoagulant drugs, the presence of rhegmatogenous retinal detachment, and dropped lens fragment. The main precipitating factor in these cases seems to be intraoperative hypotony.
To date, no significant evidence exists about the incidence, risk factors, and outcomes of delayed suprachoroidal hemorrhage after vitrectomy. Better understanding of the associations between delayed suprachoroidal hemorrhage and risk factors may enable clinicians to target high-risk patients, in an attempt to prevent this devastating complication. The aim of this study was to investigate the incidence of delayed suprachoroidal hemorrhage after pars plana vitrectomy surgery and factors associated with its development. Secondary objectives were to evaluate the interplay between risk factors and the incidence of delayed suprachoroidal hemorrhage, and to describe the anatomic and functional outcomes in these eyes.
Methods
In this retrospective multicenter cohort study, we included all consecutive patients who underwent primary pars plana vitrectomy from January 1, 2009, to December 31, 2014, at 4 tertiary vitreoretinal centers in Italy: the Eye Clinic of the University of Catania, the Santa Marta Ophthalmologic Clinic of Catania, the Eye Clinic of the University of Ancona, and the Eye Clinic of the University of Bari. The study protocol, approved by the institutional review boards of the coordinating center (University of Catania, Catania, Italy) and the other participating centers, conformed to the tenets of the Declaration of Helsinki.
Eyes that underwent pars plana vitrectomy for ocular trauma, eyes that had previous glaucoma surgery, and patients who had had intraoperative expulsive hemorrhage were not included in the study. If both eyes of the same patient had undergone vitrectomy, only 1 was randomly selected for inclusion.
Enrolled patients were divided into 2 groups on the basis of presence of delayed suprachoroidal hemorrhage, determined by all of the following criteria: (1) sudden painful loss of vision and elevated intraocular pressure (IOP); (2) shallow and/or flat anterior chamber at slit-lamp examination; and (3) dark, no serous choroidal detachment at fundus examination, confirmed by B-scan ultrasonography. All eyes that developed delayed suprachoroidal hemorrhage within 48 hours of hospitalization after the end of the vitrectomy were identified as the delayed suprachoroidal hemorrhage group. All other eyes that underwent vitrectomy in the same period, without delayed suprachoroidal hemorrhage, were considered the control group.
Patient demographics and systemic, ophthalmic, operative, and postoperative data were abstracted from the electronic medical records. For each center, 2 separate abstractors reviewed the charts of patients independently. They had been trained in the methods of chart abstraction. Definitions for key variables and all data abstraction forms were reviewed. Chart abstractors were masked to the study hypothesis.
Demographic data included age and sex. Systemic factors included presence of hypertension or diabetes, history of cerebral stroke or myocardial infarction, and use of anticoagulants/antiplatelet agents. Ophthalmic characteristics included glaucoma, axial length, previous ocular surgery, preoperative IOP, lens status, and posterior capsule status. Operative variables included indication for vitreoretinal surgery (rhegmatogenous retinal detachment, macular hole, epiretinal membrane, diabetic retinopathy, dropped lens/IOL, other indications), type of surgical procedure (vitrectomy or combined vitrectomy and phacoemulsification), size of instrument used (20, 23, or 25 gauge), use of photocoagulation (extensive photocoagulation: 360 degree/panretinal photocoagulation, for prevention of retinal redetachment, or for treatment of proliferative diabetic retinopathy and other ischemic retinal vascular diseases; localized photocoagulation: for treatment of retinal breaks, degenerative areas, or limited area of ischemia; none photocoagulation), use of cryotherapy, use of buckling, tamponade agent (air/gas or silicone oil/heavy silicone oil), sclerotomy closure (sutureless or with suture), and type of anesthesia (local or general). The postoperative variables analyzed are those that occurred after the end of the vitrectomy until the diagnosis of delayed suprachoroidal hemorrhage, and included episodes of emesis and trauma.
In all eyes with delayed suprachoroidal hemorrhage, the interval between vitrectomy and onset of delayed suprachoroidal hemorrhage, characteristics of delayed suprachoroidal hemorrhage, presence of concomitant retinal detachment, pharmacologic treatment, surgical treatment, and functional and anatomic outcomes were evaluated.
Statistical Analysis
The total number of vitrectomy procedures was identified, and the overall incidence rate was calculated. Potential risk factors were individually compared between cases and controls in univariate analyses, using χ 2 or Fisher exact tests for categorical variables and Mann-Whitney tests for quantitative variables.
Logistic regression analysis was performed to test for independence between each risk factor and delayed suprachoroidal hemorrhage and quantify risk factors for developing delayed suprachoroidal hemorrhage. Risk factors that were significant at the P < .2 level in the univariate analysis were included in the logistic regression.
We further investigated the predicting capability of the variables using a regression based on conditional inference decision trees. Risk models were developed with the use of decision-tree induction from class-labeled training records (that is, the training set was composed of records in which 1 attribute was the class label [or dependent variable] and the remaining attributes were the predictor variables; the individual records are the tuples for which the class label is known), as previously described.
Visual acuity was measured by the Snellen chart and converted to logarithm of minimal angle of resolution (logMAR) units for statistical analysis; the difference between pre– and post–delayed suprachoroidal hemorrhage logMAR units was calculated by Wilcoxon signed rank test.
All differences were considered to be statistically significant at a 5% probability level, and all reported P values are 2-sided. Statistical analysis used IBM SPSS Statistics for Windows (Version 21.0; IBM Corp, Armonk, New York, USA).
Results
A total of 4852 eyes underwent vitrectomy at 4 surgical units between January 2009 and December 2014 and were included in this analysis. The mean ± SD age was 61 ± 9 years. Patients were predominantly male (56.0%). Overall, 39 eyes presented with postoperative delayed suprachoroidal hemorrhage after vitrectomy, an incidence of 0.8%.
With regard to the indication for vitrectomy, delayed suprachoroidal hemorrhage developed in 30 of 2913 eyes with rhegmatogenous retinal detachment (1.0%), 2 of 586 eyes with macular hole (0.3%), 4 of 305 eyes with diabetic retinopathy (1.3%), 2 of 284 eyes with dropped lens/IOL (0.7%), and 1 of 274 eyes with other indications (0.4%).
Regarding surgical methods, delayed suprachoroidal hemorrhage developed in 15 of 1268 eyes with combined vitrectomy and phacoemulsification (1.2%), 11 of 1074 eyes with 20 gauge vitrectomy (1.0%), 13 of 1800 eyes with 23 gauge vitrectomy (0.7%), 15 of 1978 eyes with 25 gauge vitrectomy (0.8%), 23 of 1191 eyes with extensive photocoagulation (1.9%), 13 of 2274 eyes with localized photocoagulation (0.6%), 3 of 1387 eyes with no photocoagulation (0.2%), 28 of 2514 eyes with air/gas tamponade (1.1%), 9 of 712 eyes with silicone oil/heavy silicone oil tamponade (1.3%), none of 91 eyes with cryotherapy, none of 38 eyes with scleral buckling, and 24 of 3138 eyes with sutureless sclerotomy (0.8%).
Univariate Analysis
Univariate analysis showed that significant variables associated with increased risk of delayed suprachoroidal hemorrhage included age (67 ± 6 years in cases, 61 ± 8 years in controls; odds ratio [OR], 1.27; P < .001), hypertension (61.5% of cases, 41.2% of controls; OR, 2.29; P = .014), use of anticoagulants/antiplatelets (28.2% of cases, 15.3% of controls; OR, 2.18; P = .041), axial length (26.7 ± 0.9 in cases, 25.5 ± 1.2 in controls; OR, 2.06; P < .001), presence of rhegmatogenous retinal detachment (76.9% of cases, 59.9% of controls; OR, 2.23, P = .033), extensive intraoperative photocoagulation (59% of cases, 24.3% of controls; OR, 4.49; P = .001), air/gas tamponade (71.8% of cases, 51.7% of controls; OR, 2.38; P = .015), and postoperative emesis (46.2% of cases, 3.2% of controls; OR, 25.76; P < .001). The variables associated with reduced risk of delayed suprachoroidal hemorrhage were epiretinal membrane (0% of cases, 10.2% of controls; P = .029) and no laser treatment (7.7% of cases, 28.8% of controls; OR, 0.21; P = .002) ( Table 1 ).
| Variables | DSCH Group n = 39 | Control Group n = 4813 | P Value | OR (95% CI) |
|---|---|---|---|---|
| Demographic | ||||
| Mean ± SD age (y) | 67 ± 6 | 61 ± 8 | <.001 a | 1.27 (1.17–1.37) |
| Male sex, n (%) | 24 (61.5) | 2695 (56.0) | .521 | 1.26 (0.66–2.40) |
| Systemic | ||||
| Hypertension, n (%) | 24 (61.5) | 1982 (41.2) | .014 a | 2.29 (1.20–4.37) |
| Diabetes, n (%) | 7 (17.9) | 1309 (27.2) | .277 | 0.59 (0.26–1.33) |
| Myocardial infarction, n (%) | 1 (2.6) | 197 (4.1) | 1.000 | 0.62 (0.08–4.51) |
| Cerebral stroke, n (%) | 0 (0.0) | 63 (1.3) | 1.000 | – |
| Anticoagulants / antiplatelet agents, n (%) | 11 (28.2) | 736 (15.3) | .041 a | 2.18 (1.08–4.39) |
| Ophthalmic | ||||
| Mean ± SD axial length (mm) | 26.7 ± 0.9 | 25.5 ± 1.2 | <.001 a | 2.06 (1.63–2.59) |
| Mean ± SD preoperative IOP (mm Hg) | 13.7 ± 2.1 | 13.8 ± 2.7 | .176 | 0.98 (0.95–1.01) |
| Pseudophakic/aphakic, n (%) | 16 (41.0) | 1839 (38.2) | .742 | 1.13 (0.59–2.14) |
| Previous intraocular surgery, n (%) | 18 (46.2) | 1915 (39.8) | .417 | 1.30 (0.69–2.44) |
| Posterior capsule break, n (%) | 2 (5.1) | 72 (1.5) | .119 | 3.56 (0.84–15.05) |
| Glaucoma, n (%) | 4 (10.3) | 586 (12.2) | 1.000 | 0.82 (0.29–2.33) |
| Operative | ||||
| Rhegmatogenous retinal detachment, n (%) | 30 (76.9) | 2883 (59.9) | .033 a | 2.23 (1.06–4.71) |
| Macular hole, n (%) | 2 (5.1) | 584 (12.1) | .224 | 0.39 (0.09–1.63) |
| Epiretinal membrane, n (%) | 0 (0.0) | 489 (10.2) | .029 a | – |
| Diabetic retinopathy, n (%) | 4 (10.3) | 301 (6.3) | .306 | 1.71 (0.60–4.85) |
| Dropped lens / IOL, n (%) | 2 (5.1) | 282 (5.9) | 1.000 | 0.87 (0.21–3.62) |
| Others, n (%) | 1 (2.6) | 273 (5.7) | .724 | 0.44 (0.06–3.20) |
| Combined vitrectomy and phacoemulsification, n (%) | 15 (38.5) | 1253 (26.0) | .097 | 1.78 (0.93–3.40) |
| 20 gauge vitrectomy, n (%) | 11 (28.2) | 1063 (22.1) | .338 | 1.39 (0.69–2.79) |
| 23 gauge vitrectomy, n (%) | 13 (33.3) | 1787 (37.1) | .740 | 0.85 (0.43–1.65) |
| 25 gauge vitrectomy, n (%) | 15 (38.5) | 1963 (40.8) | .871 | 0.91 (0.47–1.73) |
| No photocoagulation, n (%) | 3 (7.7) | 1384 (28.8) | .002 a | 0.21 (0.06–0.67) |
| Localized photocoagulation, n (%) | 13 (33.3) | 2261 (47.0) | .107 | 0.56 (0.29–1.10) |
| Extensive photocoagulation, n (%) | 23 (59.0) | 1168 (24.3) | <.001 a | 4.49 (2.36–8.52) |
| Buckling, n (%) | 0 (0.0) | 38 (0.8) | 1.000 | – |
| Cryotherapy, n (%) | 0 (0.0) | 91 (1.9) | 1.000 | – |
| Air/gas tamponade, n (%) | 28 (71.8) | 2486 (51.7) | .015 a | 2.38 (1.18–4.80) |
| Silicone oil / heavy silicone oil, n (%) | 9 (23.1) | 703 (14.6) | .168 | 1.75 (0.83–3.71) |
| Eyes with sutured sclerotomies, n (%) | 15 (38.5) | 1699 (35.3) | .737 | 1.15 (0.60–2.19) |
| Eyes with unsutured sclerotomies, n (%) | 24 (61.5) | 3114 (64.7) | .737 | 0.87 (0.46–1.67) |
| General anesthesia, n (%) | 2 (5.1) | 331 (6.9) | 1.000 | 0.73 (0.18–3.05) |
| Local anesthesia, n (%) | 37 (94.9) | 4482 (93.1) | 1.000 | 1.37 (0.33–5.69) |
| Postoperative | ||||
| Emesis, n (%) | 18 (46.2) | 155 (3.2) | <.001 a | 25.76 (13.45–49.32) |
| Ocular trauma, n (%) | 0 (0) | 2 (0.0) | 1.000 | – |
Multivariable Analysis
Significant variables at univariate analysis were taken forward to multivariable analysis. A regression model was constructed for the risk of developing delayed suprachoroidal hemorrhage. The results are shown in Table 2 . Significant risk factors for developing delayed suprachoroidal hemorrhage included age (OR, 2.22; P < .001), axial length (OR, 2.57; P < .001), presence of rhegmatogenous retinal detachment (OR, 3.27; P = .005), extensive intraoperative photocoagulation (OR, 4.94; P < .001), and emesis postoperatively (OR, 24.39; P < .001).
