Purpose
To evaluate changes in the prevalence of antiplatelet (aspirin, clopidogrel) and anticoagulation (warfarin) therapy and its possible relationship to postoperative bleeding in vitreoretinal surgery (VRS) patients.
Design
Observational, retrospective case control study.
Methods
setting: University practice. study population: A total of 822 patents who underwent VRS during 3 intervals in 1994, 2004, and 2008. observation procedure: Retrospective chart review for 1994 and 2004, but contemporaneous in 2008. main outcome measures: Proportion using antiplatelets or anticoagulants, the incidence of early postoperative intraocular bleeding in patients, and clinical consequence of the hemorrhage.
Results
Thirty-one of 213 patients (14.6%) who underwent VRS in 1994, 103 of 361 patients (28.5%) in 2004, and 80 of 248 patients (32.3%) in 2008 had taken antiplatelet therapy ( P < .001). The rates of anticoagulant therapy did not vary. The incidence of bleeding was higher (20.0%) in the patients who did not suspend antiplatelets than in those who did (9.6%) ( P = .05, χ 2 test), but this difference lost statistical significance in a multivariate analysis ( P = .079). Anticoagulant was associated with intraocular hemorrhage at postoperative first day after vitrectomy ( P = .03, Fisher exact test). No reoperation or failure of the surgery was attributable to the hemorrhage in anticoagulant or antiplatelet patients.
Conclusions
Use of antiplatelet agents has increased in patients undergoing vitreoretinal surgery but probably does not increase the risk of postoperative intraocular bleeding; however, when safe to suspend even for a short time the potential risk is further reduced. Anticoagulant use was associated with a higher risk, but without serious consequences. Working with a patient’s medical doctor may allow safe suspension in some cases, which may further lower these risks.
In 2005, about one-fifth of US adults reported taking aspirin either every day or every other day, including almost half of those aged 65 and over, the age range at highest risk for cardiovascular and cerebrovascular accident. Many patients undergoing vitreoretinal surgery are in this same age range. It stands to reason that the prevalence of antiplatelet or anticoagulant use among incident vitreoretinal surgery patients would be high, and increasing. The use of warfarin in this same age group also seems to be increasing; a commensurate increased risk of hemorrhagic stroke attributed largely to anticoagulant use may have similar effect on a postoperative eye. Although perioperative hemorrhage rates with antiplatelets and anticoagulants have been studied during cataract surgery, there is little data available regarding vitreoretinal surgery. The possibility of suspending antiplatelet or anticoagulant treatment prior to any surgery in order to minimize bleeding poses difficult questions because of the potentially devastating consequences of a thromboembolic event, although a shorter-than-previously-supposed anticoagulation suspension interval may offer substantial benefits while minimizing systemic risks.
The purpose of this study was to evaluate the prevalences of antiplatelet (aspirin, clopidogrel) and anticoagulant (warfarin) therapy, and their relationship to postoperative bleeding among the patients who underwent vitreoretinal surgery during 3 sample periods over the past 15 years.
Subjects and Methods
The medical records of all patients who underwent vitreoretinal surgery by the same surgeon (W.E.S.) at the Bascom Palmer Eye Institute from January 1 through August 31, 1994, January 1 through August 31, 2004, and April 1 through August 31, 2008 were identified from a surgical database.
Preoperative data tabulated included age, gender, indication for vitreoretinal surgery, reported history of medical conditions such as hypertension and diabetes, smoking history, antiplatelet (aspirin, clopidogrel), anticoagulant (warfarin), surgical procedure, and intraocular hemorrhage at postoperative day 1. Postoperative intraocular hemorrhage was defined as hyphema, vitreous hemorrhage, subretinal hemorrhage, or choroidal hemorrhage of more than a minimal degree, sufficient to be commented in the notes. Usual, minimal hemorrhage typical of vitreoretinal surgery was not counted. The prevalence of antiplatelet and anticoagulation therapy was ascertained for each time period. Antiplatelet suspension was defined when it had been discontinued for at least 3 days preoperatively. The investigators recognized the usual medical recommendations suggest a 10- to 14-day period of suspension for antiplatelets to allow new platelets to repopulate. However, there is some evidence that a shorter suspension at least confers some reversal effect, and the 3-day interval was chosen empirically as a common threshold in this data set and reflected an actual practice among these patients whose cases were characteristically urgent rather than elective.
Intraocular hemorrhage at the first-postoperative-day examination was the primary study endpoint. The rates were compared among those with anticoagulants, with antiplatelets (with and without suspension), and without either by the χ 2 test.
To elucidate anticoagulant or antiplatelet use as independent risk factors for postoperative hemorrhage after vitreoretinal surgery, patients were analyzed for other possible risk factors in groups according to the presence (Group 1) or absence (Group 2) of postoperative intraocular hemorrhage at the first postoperative examination after vitrectomy. Independent variables evaluated for possible relation to postoperative hemorrhage were: 1) antiplatelet or anticoagulant use; 2) general characteristics (age, sex, diabetes, hypertension, smoking history); 3) procedures combined with vitrectomy (gas tamponade, scleral buckling procedure, lensectomy, and Baerveldt glaucoma implant); and 4) indications for vitrectomy, such as proliferative diabetic retinopathy (PDR) and retinal detachment (RD).
Univariate analysis (χ 2 test) was used to compare basic characteristics of Group 1 and Group 2 (SPSS for Windows, version 12.0; SPSS Inc, Chicago, Illinois, USA). For contingency tables, if more than 25% of the cells had the expected values of <5.0, we used the Fisher exact test in the SPSS software. The t test was used for 2 independent samples. Multivariate logistic regression analysis (LR backward selection) was done to determine the risk factors for postoperative hemorrhage.
Results
The study cohort comprised 822 eyes of 822 patients ( Table 1 ). There were 412 male patients (50.1%) and 410 female patients (49.9%). The mean age was 63 years (median 66, range 6-96). Seven hundred forty patients (90%) underwent pars plana vitrectomy with or without other procedures; 82 patients (10%) underwent only scleral buckling procedure. There were 213 patients (25.9%) in the 1994 group, 361 (43.9%) in the 2004 group, and 248 (30.2%) in the 2008 group. Inexplicably, the male proportion was less in the 1994 group than in other groups ( P = .002). The incidences of hypertension (44.6% [95/213], P = .036) and diabetes (26.3% [56/213], P = .032) history were also less in the 1994 group.
| 1994 | 2004 | 2008 | Total | P Value | |
|---|---|---|---|---|---|
| Number of patients | 213 | 361 | 248 | 822 | — |
| Age, years (range) | 65 (12–91) | 63 (6–94) | 63 (14–96) | 63 (6–96) | .330 |
| Male | 85 (39.9%) | 188 (52.1%) | 139 (56.0%) | 412 (50.1%) | .002 |
| Hypertension a | 44.6% | 55.8% | 52.0% | 51.7% | .036 |
| Diabetes a | 26.3% | 36.0% | 36.5% | 33.5% | .032 |
| Procedures | <.001 | ||||
| Vitrectomy | 180 | 319 | 241 | 740 | — |
| SBP only | 33 | 42 | 7 | 82 | — |
| Indication for surgery | |||||
| PDR | 30 | 53 | 44 | 127 | .481 |
| Glaucoma | 4 | 17 | 2 | 23 | .010 |
| RD | 68 | 113 | 82 | 263 | .900 |
| Other b | 111 | 178 | 120 | 409 | .709 |
| Antiplatelet use | 31 (14.6%) | 103 (28.5%) | 80 (32.3%) | 214 (26.0%) | .002 |
| Aspirin only | 31 | 74 | 63 | 168 | — |
| Clopidogrel only | 0 | 18 | 7 | 25 | — |
| Both | 0 | 11 | 10 | 21 | — |
| Warfarin use | 2 (0.9%) | 14 (3.9%) | 9 (3.6) | 25 (3.0%) | .114 |
a Twenty-three patients were missing a definitive comment regarding history of diabetes and hypertension.
b Other = retinal detachment, macular hole, epiretinal membrane, lens complication, nondiabetic vitreous hemorrhage, vitreous opacity, endophthalmitis, posterior vitreoretinopathy, etc.
The indication for vitreoretinal surgery among the 3 time intervals did not differ for PDR, RD, or others, but glaucoma-associated procedures were more prevalent in the 2004 group ( P =.010) ( Table 1 ).
Antiplatelet or Anticoagulant Use in Different Time Intervals
Among all 822 patients, 214 (26.0%) had taken antiplatelet agents including aspirin only (168, 20.4%), clopidogrel only (25, 3.0%), or both (21, 2.6%); 145 (68%) had suspended their antiplatelets for at least 3 days preoperatively ( Table 1 ). For the 1994 group, 31 of 213 patients (14.6%) had taken antiplatelets (clopidogrel had not been introduced yet); for the 2004 group, 103 of 361 patients (28.5%) took antiplatelet agents (74 with aspirin only, 18 with clopidogrel only, 11 with both) ( P = .001); for the 2008 group, 80 of 248 patients (32.3%) took antiplatelet agents (63 with aspirin only, 7 with clopidogrel only, 10 with both) ( P = .001). The frequency of antiplatelet use increased from 1994 to 2004 and 2008 ( P < .002), but the prevalence did not change from 2004 to 2008 ( P = .324). The incidence of antiplatelet use was not different during the 3 time periods for patients whose vitreoretinal surgery indication was PDR or glaucoma ( Table 2 ), but it was higher for an indication of RD ( P = .006).
| Indication for Surgery | 1994 Antiplatelet/Total | 2004 Antiplatelet/Total | 2008 Antiplatelet/Total | P Value a |
|---|---|---|---|---|
| PDR | 4/30 (13.3%) | 7/53 (13.2%) | 8/44 (18.2%) | .760 |
| Glaucoma | 0/4 (0%) | 1/17 (5.9%) | 1/2 (50%) | .194 b |
| Retinal detachment | 2/68 (2.9%) | 17/113 (15.0%) | 17/82 (20.7%) | .006 |
| Other | 13/111 (11.7%) | 29/178 (16.3) | 25/120 (20.8%) | .173 |
| Total | 31/213 (14.6%) | 103/361 (28.5%) | 80/248 (32.3%) | .002 |
The frequency of anticoagulant use did not differ by time interval. Twenty-five of 822 patients (3%) had taken anticoagulants, including 2 of 213 patients (0.9%) in 1994, 14 of 361 patients (3.9%) in 2004, and 9 of 248 patients (3.6%) in 2008 ( P = .114).
The indication for antiplatelet use was documented in 54.8% (17/31) in 1994 and 53.4% (55/103) in 2004, but only 21 of 80 patients (26.3%) in 2008 ( P < .001)—possibly because of increased familiarity by the screeners with its use ( Table 2 ). The indication for antiplatelets was ischemic cardiovascular or cerebrovascular disease (83, 38.8%), arrhythmia (9, 4.2%), valvular heart disease (1, 0.5%), and unknown (121, 56.5%) for the combined group. The indication for anticoagulant use was documented in all patients and included cardiovascular or cerebrovascular disease (18, 72%), arrhythmia (6, 24%), and valvular heart disease (1, 4%).
Among patients taking antiplatelets, they were suspended for at least 3 days before surgery in 18 of 31 patients (58.1%) in the 1994 group, in 72 of 103 patients (69.9%) in the 2004 group, and in 55 of 80 patients (68.7%) in the 2008 group ( P = .452, χ 2 test).
Anticoagulants were not suspended for 5 days or more in any patients, but they were suspended for smaller intervals in 18 of 25 patients. These numbers were insufficient to yield any meaningful statistical analyses. Of the 21 patients who had taken anticoagulants, the international normalized ratio (INR) for prothrombin time was available in 11 patients; the median was 1.25 (range, 0.80–3.15). Postoperative intraocular hemorrhage occurred in 2 of 11 patients whose INR was checked. Their INRs were 1.21 and 3.15. Again, no meaningful statistical analyses could be derived.
Incidence of Postoperative Bleeding Versus Antiplatelet or Anticoagulation Use
Among the 822 patients 93 eyes had bleeding at the first postoperative examination attributable to vitreous hemorrhage (66), hyphema (17), choroidal hemorrhage (6), and subretinal hemorrhage (4) ( Table 3 ). Twenty-five patients had taken anticoagulants, 214 antiplatelets, and 588 no anticoagulants or antiplatelets; 5 patients used both anticoagulants and antiplatelets, and 69 patients had taken antiplatelets without suspension for at least 3 days prior to surgery. The rate of bleeding for the subgroup that did not suspend antiplatelets (17% ,12/69) was higher than for the group that did suspend antiplatelets (9.6%, 14/145) ( P = .05) but was similar to the group that had not used antiplatelets at all (10%, 61/588) ( P = .079). In a multivariate model, however, the use of antiplatelets without suspension was not statistically significant ( Supplemental Tables 1 and 2 , available at AJO.com ). The rate of bleeding in patients with anticoagulants (28%, 7/25), however, was higher than in those without antiplatelets or anticoagulants (10%, 61/588) ( P = 0.014, χ 2 test) ( Supplemental Figure , available at AJO.com ). The rates of bleeding among patients using aspirin only (18%, 98/51), clopidogrel only (18%, 9/51), and both (13%, 1/8) were similar.
| After Vitrectomy n = 740 | After SBP n = 82 | P Value a | Total n = 822 | |
|---|---|---|---|---|
| Postoperative hemorrhage | 90 (12.2%) | 3 (3.7%) | .021 | 93 |
| Vitreous hemorrhage | 66 | 0 | — | 66 |
| Hyphema | 17 | 0 | — | 17 |
| Choroidal hemorrhage | 5 | 1 | — | 6 |
| Subretinal hemorrhage | 2 | 2 | — | 4 |
None of the 214 patients taking antiplatelets or anticoagulants suffered an operation failure or reoperation attributable to the hemorrhage. However, 4 of 61 patients not taking antiplatelets or anticoagulants who experienced bleeding had reoperation attributable to the hemorrhage, including 1 whose operations failed to restore vision. Unanticipated ultrasound examinations were performed for 3 of the 61 bleeds (5%) not on anticoagulants or antiplatelets, compared to 1 of 7 (4%) on anticoagulants and 3 of 25 (12%) on antiplatelets ( P = .313).
No thromboembolic events were tabulated in any of the patients in the study during the follow-up interval.
Risk Factors for Postoperative Bleeding
Ninety-three patients were included in Group 1 (with bleeding) and 729 in Group 2 (without bleeding). A univariate analysis of risk factors was carried out for general characteristics, indications for surgery, and various components of the vitrectomy ( Table 4 ). Early postoperative bleeding occurred more frequently in male than female patients ( P = .006), in patients with a smoking history ( P =.005) and in diabetic patients ( P < .001), but did not differ by age ( P = .366) or hypertension ( P = .126). Postoperative intraocular bleeding was more common in patients with PDR ( P < .001) and glaucoma ( P < .001) than among the other diagnostic subgroups.
| Variable | Early Postoperative Hemorrhage | P Value a | |
|---|---|---|---|
| Yes (Group 1) (n = 93) | No (Group 2) (n = 729) | ||
| Gender | |||
| Male | 59 (63.4%) | 353 (48.4%) | .006 |
| Female | 34 (36.6%) | 376 (51.6%) | — |
| Age | 61.86 ± 13.8 | 63.4 ± 15.7 | .366 b |
| Smoking | 40 (46.0%) | 212 (30.9%) | .005 |
| Diabetes | 51 (54.8%) | 217 (30.7%) | <.001 |
| Hypertension | 55 (59.1%) | 358 (50.7%) | .126 |
| Indication for surgery | |||
| PDR | 35 | 92 | <.001 |
| Glaucoma | 10 | 13 | <.001 c |
| Retinal detachment | 22 | 241 | .067 |
| Other | 26 | 383 | <.001 |
| Procedure | |||
| SBP only | 3 | 79 | .021 |
| PPV | 79 | 650 | — |
| With gas tamponade | 15 | 267 | <.001 |
| With SBP | 8 | 70 | .757 |
| With lensectomy | 15 | 47 | .001 |
| With BGI | 7 | 8 | .001 c |
| Antiplatelet use | 26 (28.0%) | 188 (25.8%) | .654 |
| Antiplatelet without suspension | 12 (12.9%) | 57 (7.8%) | .096 |
| Antiplatelet with suspension | 14 (15.1%) | 131 (18.0%) | .487 |
| Anticoagulant use | 7 (7.5%) | 18 (2.5%) | .016 c |
| No antiplatelet or anticoagulant | 61 (65.6%) | 527 (72.3%) | .178 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
